In addition to these nine countries that possess nuclear weapons of their own, U.S. nuclear weapons are reportedly located in six other countries—one other nuclear weapon state(the United Kingdom), and five non-nuclear-weapon states (Germany, the Netherlands, Belgium, Italy, and Turkey). [2]

World stockpiles of separated plutonium and HEU, the essential ingredients of nuclear weapons, amount to well over 2,300 tons—enough to manufacture over 200,000 nuclear weapons. [3]Neither of these materials occurs in significant quantities in nature; these stockpiles of weapons and materials have all been consciously produced by human beings in the six decades of the nuclear age.

Unlike nuclear weapons, separated plutonium and HEU have both military and civilian uses.  A number of countries reprocess plutonium from spent fuel and recycle it as plutonium-uranium mixed oxide (MOX) fuel in civilian reactors, resulting in the processing, transport, and use of many tons of weapons-usable separated plutonium every year.  In recent years, use of the separated plutonium as fuel has not kept pace with reprocessing, with the result that as of the end of 2003, nearly 240 tons of separated, weapons-usable plutonium existed in civilian stockpiles worldwide—a figure that will soon surpass the total amount of separated plutonium in all the world's nuclear weapon stockpiles. [4] HEU is no longer used in civilian power reactors (with a couple of exceptions), but remains widely used in civilian research reactors (as well as for medical isotope production, in naval and icebreaker reactors, as spike fuel in plutonium and tritium production reactors, and for some other purposes).  Roughly 140 research reactors in more than 40 countries continue to operate with HEU as their fuel. [5] Some of these do not have enough nuclear material on-site for a bomb, but many do— as do many associated facilities, such as fuel fabrication plants.  All told, there are an estimated 128 research reactors or associated facilities that possess at least 20 kilograms of HEU. [6] Of these, 41 are fuel facilities rather than research reactors themselves. [7] There are an estimated 65 tons of HEU in civilian use worldwide. [8] As a result, while nearly half of the estimated world stockpile of nearly 490 tons of separated plutonium at the end of 2003 was civilian, only about 3% of the estimated world stockpile of HEU was civilian. 

The IAEA does not safeguard military nuclear material, and nuclear weapon states are not required to place their nuclear materials under IAEA safeguards (though a small amount of material in these states, particularly in the United States, is under safeguards under voluntary offer agreements, and French and British civilian material is under Euratom safeguards, integrated with the IAEA). Hence, as of the end of 2003 less than 2% of the world's estimated HEU stockpile was under safeguards (representing 31.8 tons of HEU, of which 10 tons was excess U.S. HEU), and only about 17% of the world's separated plutonium was under safeguards (representing 89.6 tons of separated plutonium outside of reactor cores, some 80% of that in Britain and France). [9]

Because they have both military and civilian uses, these materials are much more broadly distributed than nuclear weapons are.  Separated plutonium or HEU exist in hundreds of buildings in more than 40 countries.  There are ten countries with two metric tons or more of separated plutonium or HEU, including all of the five NPT nuclear weapon states, India (a non-NPT state), Germany, Japan, Belgium, and Kazakhstan.  Thus there are at least three non-nuclear-weapon states under the NPT with enough weapons-usable nuclear material on their soil for hundreds of nuclear weapons. [10]
See Table 2: Countries With ≥2 Tons of Weapons-Usable Nuclear Material.

Beyond the countries with tons of weapons-usable nuclear material, there are roughly 26 additional countries with "Category I" quantities of weapons-usable nuclear material—that is, enough material that under international standards, the highest levels of security are required (this applies to more than 5 kilograms of U-235 contained in HEU, or more than 2 kilograms of plutonium.)  This includes the three other non-NPT states and 23 additional NPT non-nuclear-weapon states.  Of these 26, seven are developing countries and nine are transition countries (that is, former communist countries).  Thus, nuclear weapons or enough nuclear material to pose a serious concern exist in a total of some 36 countries.
See Table 3: Other Countries With Category I Quantities of Weapons-Usable Material.

Security for these materials in all of these countries must be effective enough to ensure that plausible terrorist and criminal threats, both from insiders and outsiders, can be reliably defeated.

Beyond these countries, quantities of separated plutonium or HEU in the range of roughly one to a few kilograms exist in an additional 13 countries.  All of these are NPT non-nuclear weapon states; seven are developing countries, two are transition countries. 
See Table 4: Countries With Kilogram-Range Quantities of Weapons-Usable Material. 
Hence, quantities in the range of a kilogram or more of HEU or separated plutonium exist in roughly 49 countries. Because official information on the stocks of HEU in these different countries is generally not publicly available, these tables are based on partial information and judgment; it may be that a few countries should be added, subtracted, or moved from one table to the other.

The countries with either nuclear weapons or substantial stockpiles of nuclear materials, shown in Tables 1 and 2, generally have between a dozen and hundreds of buildings where their nuclear stockpiles reside.  The countries without nuclear weapons and with between a kilogram and two tons of weapons-usable nuclear material on their soil (shown in Tables 3 and 4) typically have only one or two buildings with weapons-usable nuclear material, though a small number have up to half a dozen such buildings.  No complete estimate of the number of buildings worldwide with a kilogram or more (or a Category I quantity or more) of weapons-usable nuclear material exists; that figure is likely to be over 1,000 buildings, but is certainly less than 3,000.

Transport of Stockpiles

Nuclear warheads and weapons-usable materials must be adequately secured not only while they are at fixed facilities, but also while they are being transported — between buildings within sites, between sites, and between countries.  Indeed, transport is the stage of these items' life cycle that is most vulnerable to overt, forcible theft, as when these items are being shipped from place to place, it is impossible to provide the multiple layers of detection and delay that can be put in place at a fixed site.  This problem is typically addressed with measures such as armed guards accompanying the transports, vehicles with special protection against hijack and sabotage, secrecy concerning the schedule and route of the transports, and continuous or frequent tracking of the transport en route.

Weapons.  The scale and frequency of transport, particularly from site to site within countries, is great.  Hundreds of nuclear warheads are transported from deployment sites to warhead storage and assembly/disassembly facilities, or from such facilities back to deployment sites, each year, in both Russia and the United States—and presumably, to a lesser extent, in other countries with nuclear weapons.  In Russia, for example, the U.S. Cooperative Threat Reduction (CTR) program has been planning to pay for roughly 70 shipments per year of nuclear warheads to dismantlement and storage sites, carrying 20-30 warheads each [11] — in addition to however many shipments take place for operational purposes (which are not paid for by the United States). (See our discussion on Warhead Security.)  In the United States, within DOE alone, the Secure Transportation Asset program carries out nearly 100 secure transports of either nuclear warheads or weapons-usable nuclear material a year, at an annual cost that is now in the range of $140 million per year. [12] That does not include Department of Defense transport of nuclear weapons and materials, or private transport of nuclear materials.

HEU.  Transport of military HEU takes place on a similarly massive scale, as nuclear weapons are dismantled, HEU is shipped from dismantlement facilities to HEU storage facilities, and, in Russia and the United States, excess HEU is shipped to other facilities for blending to LEU.  The 30 tons of HEU Russia blends to LEU each year for sale to the United States (see our page on the U.S.-Russian HEU Purchase Agreement) is shipped from facility to facility and back again over thousands of kilometers of rail, in scores of annual shipments, representing probably the largest annual transport of weapons-usable nuclear material in the world (if measured in ton-kilometers traveled). [13] The scale of shipments of civil HEU is small by comparison, though the hundreds of kilograms of HEU which are shipped each year for fuel for research reactors and as targets for medical isotope production — primarily within the United States and Russia, but also internationally — also pose proliferation risks that must be addressed. [14]

Plutonium.  Transport of military plutonium currently occurs at a much smaller scale than transport of military HEU.  In the United States, and apparently to a significant degree in Russia as well, plutonium from dismantled weapons is stored at the dismantlement sites, rather than being transported elsewhere for storage, and disposition of excess plutonium — which will lead to this material being transported to processing and fuel fabrication sites, and then in the form of fabricated fuel to reactor sites — has not yet gotten underway.  Large quantities of weapons-usable separated plutonium are transported every year in the civil sector, however, as some 20 tons of plutonium is reprocessed from spent fuel and some 10 tons of that is fabricated into fuel for use in nuclear reactors.  By one estimate, roughly 100 commercial plutonium shipments occur per year, most of which contain over 100 kilograms of weapons-usable plutonium in a single shipment. [15] In France in particular, which has the world's most active plutonium recycling plutonium, many tons of plutonium separated at the La Hague reprocessing plant each year travel by scores of truck shipments, as plutonium oxide, to the fuel fabrication facility at Marcoule; once fabricated into fuel elements, this plutonium is then shipped to numerous reactors both in France and in other countries. [16] Plutonium separated by reprocessing in Britain is stored at the reprocessing site at Sellafield, without transport, and under current plans much of this plutonium will be fabricated into fuel at the MOX plant at the same site — at which point there will begin to be major shipments of plutonium in fabricated fuel every year from Britain to other countries.  Occasionally, substantial shipments of separated plutonium are shipped across the oceans, as when separated plutonium from reprocessing is returned to Japan from France and Britain, or in the recent case when tens of kilograms of weapons plutonium were shipped from the United States to France for fabrication into MOX lead test assemblies for use in a U.S. reactor. The adequacy of security for nuclear material transports around the world has been a subject of controversy for many years; the fact is that it is extraordinarily difficult to provide the same level of security for items in transport as they can have at large fixed sites. [17]

Changing Size of Stockpiles

Both the size of the global stockpiles of nuclear warheads and weapons-usable nuclear material and their distribution are changing over time — in somewhat different directions.

The total number of nuclear weapons in the world has been declining for over a decade, as the United States and Russia are believed to have dismantled many thousands of nuclear weapons since the 1980s.  This decline may be slowing substantially, however.  While in some years in the 1990s, the United States dismantled as many as 1,800 warheads in a single year, in recent years it appears that this figure has been in the range of 0-300 warheads. [18] The United States has, however, announced a substantial reduction in the planned stockpile of nuclear weapons, which may lead to the dismantlement of some 4,000 weapons by 2012 — though it appears to plan to maintain some 6,000 nuclear warheads indefinitely thereafter. [19] In Russia, some estimates in the 1990s similarly suggested a dismantlement rate in the range of 2,000 a year or even more.  But in recent years, Russia has closed two of its four weapons assembly and disassembly facilities, suggesting that it now foresees a significantly lower rate of dismantlement in the future (though the two facilities closed had modest capacities compared to the two that remain open). [20] Like the United States, however, it appears that Russia still has some thousands of warheads that are not currently needed either for operational stockpiles or for reserves, which may be dismantled over the next decade. [21]

British warhead stockpiles are also declining, and it appears that French stockpiles have been shrinking as well.  Chinese stockpiles are expected to increase modestly over the next decade, and India, Pakistan, and India are believed to be continuing to produce small numbers of warheads; though the fate of North Korea's nuclear program will depend in part on whether it lives up to its September 2007 pledge to "disable their nuclear programmes by the end of this year 2007." [22]

Warheads.  The global distribution of nuclear warheads has also declined in recent years. With the collapse of the Soviet Union and the 1991-1992 presidential nuclear initiatives, all Soviet nuclear weapons were removed from Eastern Europe, from surface ships, and from the non-Russian states of the former Soviet Union during the 1990s.  Similarly, U.S. nuclear weapons were removed from surface ships and from South Korea in the 1990s. The number of states that possess nuclear weapons of their own (nine) is the same in 2007 as it was 20 years before, as South Africa became the first and only state to completely dismantle a nuclear weapon stockpile that it owned and had full control over — but North Korea added itself to the list of states with nuclear weapons. Trends over the next 20 years are difficult to predict; it remains possible that the current number of states with nuclear weapons will remain stable or even decline (if international efforts succeed in rolling back North Korea's nuclear program, ensuring that Iran does not develop nuclear weapons and convincing other states not to follow the nuclear weapons route); but it is also possible that the number of states with nuclear weapons could increase significantly, with both North Korea and Iran becoming full-fledged nuclear powers and a number of other states subsequently choosing to follow the same path.

HEU.  Like warhead stockpiles, global stockpiles of military HEU have been falling for more than a decade.  All of the five NPT nuclear-weapon states have stopped production of HEU for weapons, and the United States and Russia have each declared substantial quantities of military HEU as excess to their military needs.  Under the U.S.-Russian HEU Purchase Agreement, Russia blends 30 tons of excess military HEU each year to LEU for sale to the United States, a program that is expected to continue until 2013, at which point 500 tons of HEU will have been destroyed.  Some 300 tons of HEU have been destroyed in this effort to date. [23] Roughly 60 tons of HEU have been destroyed in the slower-paced U.S. HEU disposition program, which ultimately plans to destroy or dispose of 174 tons of excess U.S. HEU.  Pakistan is believed to continue to produce military HEU, though on a small scale compared to the United States and Russia; India is thought also to have modest military HEU production underway, though plutonium is the primary focus of its military nuclear material production program.  North Korea was reported to be endeavoring to establish a military HEU production capability, but U.S. intelligence assesses that it is still some years away from acquiring such a capability.  Iran is working to establish a large-scale enrichment facility which it insists is for solely peaceful purposes; others are concerned that this facility or others established covertly might be turned to military purposes.  Civil HEU stockpiles have been growing modestly, as the pace at which research reactors discharge irradiated HEU and load more has been greater than the pace at which this HEU has been blended or disposed of, but the establishment of the Global Threat Reduction Initiative has accelerated removal and disposition of civil HEU, which could reverse this trend.  Civil HEU stockpiles remain tiny by comparison to military stockpiles.

Plutonium.  Global military plutonium stockpiles are nearly static.  All of the five NPT nuclear-weapon states have stopped producing plutonium for use in weapons — though in Russia, three plutonium production reactors continue to operate, churning out some 1.2 tons of plutonium a year, because they also provide essential heat and power for nearby Siberian communities.  India, Israel, and North Korea are believed to continue small-scale production of military plutonium, and such production has recently begun in Pakistan as well, complementing Pakistan's primary focus on HEU.  As noted above, disposition of excess military plutonium has not yet gotten underway.

Civil plutonium stockpiles, by contrast, continue to increase dramatically.  Every year, nuclear power plants around the world discharge some 8,000 tons of spent fuel, containing some 80 tons of plutonium.  Roughly one-quarter of this fuel is reprocessed each year, yielding some 20 tons of separated plutonium.  Only about half of that plutonium separated by reprocessing is fabricated into fuel each year, with the remainder remaining in storage.  Hence, the global stockpile of separated, weapons-usable civilian plutonium increases by roughly 10 tons each year.

Changes in the Distribution Pattern.  The global distribution of separated plutonium and HEU is changing only slowly.  All of the countries with substantial stockpiles, shown in Table 2, have had stockpiles for decades.  Essentially all of the countries with smaller stockpiles, on Tables 3 and 4, have had at least modest stockpiles of weapons-usable nuclear material for decades.  With respect to HEU, the trend is toward fewer and fewer countries having stockpiles, as the U.S. and Russian efforts to convince countries to send back the HEU they exported gain momentum. [24] (With respect to separated plutonium, the global distribution is likely to be static or nearly so for some time to come, as few additional countries are interested in pursuing a plutonium fuel cycle, but those who have separated plutonium on their soil are finding it hard to get rid of (though Switzerland is one example of a country that has burned all or nearly all of the separated plutonium it owned as MOX fuel).

Widely Varying Nuclear Security

Those seeking material for a nuclear bomb will go wherever it is easiest to steal, or buy it from anyone willing to sell.  Thus, security for bomb material is only as good as its weakest link.  Insecure nuclear bomb material anywhere is a threat to everyone, everywhere.  Yet today, there are no binding international standards for how well nuclear weapons and materials should be secured. Nuclear security levels are left to the discretion of each of the dozens of states that possess such stockpiles, with the result that security for stocks of potential nuclear weapons materials varies enormously, from excellent to appalling.

It is important to understand that the nuclear Nonproliferation Treaty (NPT) does not contain any provisions requiring states to secure nuclear material from theft. [25] Similarly, the IAEA safeguards system is designed only to verify that states have not diverted nuclear material for nuclear explosives, not to protect material from theft or even to confirm that the state that owns the material is providing adequate protection. [26] Indeed, because of the long times between inspections at many sites, the IAEA would not typically be able to detect that a theft had occurred until days, weeks, or months after the fact.  In any case, as noted above, some 90% of the world's separated plutonium and HEU is not under either IAEA or Euratom safeguards.

There is now a legally binding U.N. Security Council resolution requiring all states to provide "appropriate effective" security for any nuclear stockpiles they may have — but no one has yet defined what the essential elements of an effective system required by this resolution might be. [27] A negotiated amendment to the Physical Protection Convention will create at least some very general requirements for security for nuclear stockpiles but the convention does not apply to military stockpiles; the rules it sets are extraordinarily general (specifying, for example, that countries should set and enforce rules for how secure their nuclear facilities should be, but not what those rules should say); the amendment will not enter into force for years to come; and as of the fall of 2007 only a few countries had signed and ratified the amendment. [28] There is also a 2005 International Convention on the Suppression of Acts of Nuclear Terrorism, which entered into force in July 2007.  It focuses primarily on requiring that countries pass laws criminalizing the various acts involved in nuclear terrorism; it does require parties to "make every effort" to put in place "appropriate" security for their nuclear stockpiles, but does not further define what that means. [29]

IAEA recommendations provide the most specific international standards for nuclear security that now exist, but even these are quite vague: they specify, for example, that significant amounts of weapons-usable nuclear material should be stored in a place with a fence and intrusion detectors, but they say nothing about how strong the fence should be or how good the intrusion detectors should be. They recommend 24-hour guards, but do not require that they be armed, and say nothing about how numerous or well-equipped or well-trained they should be. They recommend that states establish a "design basis threat" that their facilities with significant amounts of weapons-usable material be required to defend against — but they do not say anything about what that threat should be. [30] Most states try to ensure that their facilities meet the IAEA recommendations, and many have agreements with nuclear suppliers that require them to do so.

A number of major nuclear suppliers, including the United States, have adopted policies or laws that require countries they supply to meet some requirements for physical protection for the supplied nuclear material. The United States, in particular, is required by law to ensure that recipient countries meet adequate physical protection standards and has nuclear supply agreements under which it has provided HEU to scores of countries around the world.  Its nuclear supply agreements with foreign countries typically require that the recipient country provide "levels of physical protection" for the supplied nuclear material "at least equivalent" to those in the IAEA recommendations. [31] The Nuclear Suppliers' Group (NSG), a cartel of the major nuclear suppliers, has agreed that all of its participants will require that recipients of major nuclear exports meet at least a more general set of physical protection criteria; these refer to the IAEA recommendations, but only as a "useful basis for guiding" recipient states in designing their physical protection systems, not as a requirement. [32] Similarly, a number of states have entered into agreements in other contexts that require certain levels of physical protection—in some cases to implement the IAEA recommendations. These include the U.S.-Russian Highly Enriched Uranium Purchase Agreement and in the African Nuclear Weapon Free Zone. [33]

Information on the specific measures taken to secure nuclear stockpiles around the world is typically kept secret; to keep potential terrorists and thieves guessing what security measures they may be up against at a particular site.  Hence it is effectively impossible, in an unclassified publication, to put together a complete picture of security for nuclear weapons and materials around the world.  It is a frightening fact that no one, today—not the U.S. government, not the International Atomic Energy Agency, not any other government or organization, as far as is known—has such a complete picture: while a great deal is known about the risks at some particular sites, no one knows for sure which sites, judged on a global basis, pose the highest risks and should be the highest priorities for policy steps to reduce the risks. [34]

Nevertheless, from the unclassified information that is available, it is clear that security arrangements vary widely from one country and facility to another. [35] In a troubling number of cases these arrangements would likely not be sufficient to deal with either a well-planned insider theft or an attack by a significant number of well-armed and well-trained outsiders.  Until the 9/11 attacks, for example, several countries did not require any armed guards at nuclear facilities—including Japan, which has tons of weapons-usable separated plutonium and hundreds of kilograms of HEU metal on its soil, enough material for many hundreds of nuclear weapons, and was the nation where the Aum Shinrikyo terror cult was working actively to get nuclear weapons and the materials to make them. [36] (Since the 2001 attacks, Japan has posted armed units of its national police to guard nuclear facilities—but these are apparently not fully integrated with the security plans at the sites, and many of them patrol at the perimeters, where they look impressive but would be vulnerable to being shot in the opening moments of an attack. [37])

Many countries have not defined in regulations or other rules any particular threat that nuclear security systems had to be able to defeat (known as the "design basis threat," or DBT, because it is the threat that is the basis for designing the security system); they relied instead on setting rules regarding how high fences should be, what types of locks and vaults should be provided, and the like.  Many experts believe that having such rules requiring a particular level of performance from the security system, rather than a compliance-based approach where rules simply require that particular technologies and procedures be in place, is crucial to good security.  As one U.S. expert put it, "if you don't have a DBT, you don't have good security." [38]

A review of presentations about their approaches to physical protection made by 19 countries at conferences in the late 1990s noted wide variations in their nuclear security approaches and practices.  For example, 12 of the 19 reported that they perceived a threat of insider theft and took measures to address that problem, six provided no information at all on the insider problem, and one country insisted that it faced no threat from insiders.  Only 11 of the 19 reported that they required facilities to protect against sabotage, as well as against theft of nuclear material. [39] In responses to a detailed survey on nuclear security practices prepared by researchers at Stanford University, five of the six respondents said they had a DBT in place, but two of the six said they did not take into account any risk of an attack by terrorists in their DBT; three of the six said their DBT did not include dangers from insiders (either for theft or for sabotage); none of the six reported having made any provision to deal with the threat of sabotage by a large truck bomb dispersing radioactive material beyond the protected area; two of the six did not require armed guards to protect areas with weapons-usable nuclear material; most required that when operations were done in an area with weapons-usable nuclear material at least two persons had to be present ("two-man rule"), but "that requirement was administered in quite different ways and in some cases not followed." [40]

Too little data is publicly available to provide a detailed country-by-country review of nuclear security arrangements in each of the countries with notable quantities of nuclear material.  A review of the information that is publicly available concerning security levels at different sites in different countries, quantities and qualities of material at those sites, and the threats that security systems in different countries must face suggests that as of 2007, the most urgent risks of nuclear theft existed in Russia; at research reactors fueled with HEU around the world; and in Pakistan. 

Nuclear Security in Russia—Yesterday and Today

The breakup of the Soviet Union in 1991 created a danger unprecedented in human history—the collapse of an empire armed with tens of thousands of nuclear weapons and enough nuclear material for tens of thousands more.  The world has been extraordinarily lucky that this collapse involved so little violence and that a horrifying outpouring of weapons of mass destruction and related materials and technologies did not occur. Substantial progress has been made in the years since the Soviet collapse, and Russia today is a very different country than Russia in the mid-1990s.  But important risks remain, and urgent action is needed to address them. (See much more on this topic in The Threat in Russia and the NIS.)

The Threat From Research Reactor Fuel

Roughly 140 research reactors, in more than 30 countries around the world, still use HEU as their fuel.  A variety of related civilian sites also have HEU on-site, including fuel fabrication and processing facilities, shut-down or converted reactors from which the HEU has not yet been removed, and reactors using HEU targets to produce medical isotopes, among others.  Many of these facilities do not have enough HEU on-site for a bomb, but a surprising number of facilities do have enough.  In November 2004, the U.S. Government Accountability Office reported that a Department of Energy study concluded that there are 128 nuclear research reactors or associated facilities around the world with 20 kilograms of HEU or more. [41] (Of these, 41 are fuel facilities rather than research reactors themselves. [42]) Moreover, one cannot rule out the possibility of terrorists stealing material from more than one facility, each of which might have less than the amount required for a bomb; the possibility of simultaneous attacks is highlighted by the simultaneous al Qaeda attacks on the U.S. embassies in Kenya and Tanzania in 1998.  The potential use of research reactor HEU in nuclear weapons is not just a hypothetical concern: as discussed in our page on the Demand for Black Market Nuclear Material, Iraq, in its "crash program" to make one nuclear bomb as quickly as possible after its invasion of Kuwait, planned to use both fresh and irradiated HEU from its research reactors. [43]

Most civilian research reactors have only modest security—in many cases, no more than a night watchman and a chain-link fence even when enough fresh or irradiated HEU for a bomb is present. [44]  Some are located on university campuses, where providing serious security against terrorist attack would be virtually impossible—and where the operators are often partly students, who cycle through frequently, making it extraordinarily difficult to provide serious checks of potential insider thieves.  Many research reactors were built 30-40 years ago, in the heyday of nuclear energy; many have since fallen on hard times and have few resources to continue safe operation or to pay for substantial security measures.  The research reactor in the Congo, attempting to operate in the midst of a civil war, grinding poverty, and endemic corruption, is emblematic of the broader problem (though its fuel is just below the 20% line that defines HEU): fuel stolen from that reactor turned up in the hands of the Italian mafia. [45]

Even in the United States, which has some of the most stringent nuclear security rules in the world, research reactors regulated by the Nuclear Regulatory Commission (NRC) are exempted from the requirement that facilities with more than 5 kilograms of U-235 in HEU emitting less than 100 rads per hour at one meter must have sufficient armed guards, fences, and other security measures in place to defeat theft attempts by either an insider or groups of armed outsider attackers. [46] In mid-2005 an investigation by ABC News documented conditions ranging from sleeping guards to security doors propped open with books at essentially all of the 26 U.S. university-based research reactors, including those fueled by HEU. [47]

Given these security conditions, it would not be difficult for attackers to break in and remove large quantities of HEU from a research reactor; insiders might also be able to remove such material.  Unlike the huge, massive fuel assemblies used in nuclear power reactors, fuel for research reactors is typically in fuel elements that are small and easy to handle — typically less than a meter long, several centimeters across, and weighing a few kilograms.  In most cases, a thief could easily put several fuel elements at a time into a backpack, to be carried out to a waiting vehicle.

In general, the HEU in these fuel elements would require some processing before it could be used in a bomb—but the kind of processing required is reasonably straightforward, and all the details of the necessary processes are published in the open literature.  While there is a broad range of different types of research reactor fuel, a very typical fuel is a mixture of uranium and aluminum, with aluminum cladding.  To separate out the uranium from the aluminum, such fuel could be cut into pieces, dissolved in acid, and the uranium separated from the resulting solution by well-known processes.  Converting the chemical forms of uranium that would be recovered by these means to metal would also involve straightforward processes, all of which are published in the open literature.  As one analysis put it, separating the uranium from research reactor fuel can be done "using commonly available equipment...all readily available in countries with even very modest chemical industries.…  [A]ll process chemistry data are published." [48] It is very likely that a terrorist group with the level of technical expertise required to make a nuclear bomb from HEU metal would also be able to solve the challenge of getting HEU metal from research reactor fuel.

It is important to understand that "spent" research reactor fuel also poses a serious proliferation threat.  First, irradiated research reactor fuels typically remain very highly enriched: many fresh research reactor fuels are 90% enriched, and are still more than 80% enriched after irradiation. [49] (The bomb that incinerated the Japanese city of Hiroshima was made from 80% enriched uranium. [50] )

Second, most of these fuel elements are not radioactive enough to prevent them from being stolen and processed for bomb material.  Fuel that emits more than 100 rem/hour at 1 meter is considered "self-protecting" under international standards, meaning that it is thought to be too radioactive for thieves to plausibly steal.  This standard should be reconsidered, for in the case of suicidal terrorists who do not care about increasing their chance of cancer years afterward, 100 rem/hour would provide little deterrent. [51] But in any case, most irradiated HEU research reactor fuel in the world does not meet this standard.  Because the fuel elements are small, are not irradiated to the same power densities as power reactor fuel, and in many cases have been sitting in pools cooling for decades, most of this material could be stolen almost as easily as the fresh material could be.

Third, because of the very modest level of radioactivity, for terrorists who do not care about their long-term cancer risks, getting the uranium out of this material for use in a bomb would be little more difficult than getting the uranium out of fresh, unirradiated fuel.  The same chemical processes described above could be used.  Thus, kilogram for kilogram, irradiated research reactor fuel poses only a modestly lower proliferation danger than fresh research reactor fuel—and there is far more irradiated HEU fuel at poorly secured reactor sites around the world than there is fresh fuel. [52] The danger posed by research reactor spent fuel stands in stark contrast to the modest theft threat posed by nuclear power reactor spent fuel assemblies, which are huge, heavy, and intensely radioactive, making them quite difficult to steal and process.

Security of Pakistan's Stockpile

Pakistan has a relatively modest nuclear stockpile, which is thought to be distributed among only a small number of locations.  Pakistan has sites where nuclear weapons exist (reportedly stored in partially disassembled form [53]), and sites with HEU or separated plutonium (particularly the main HEU production facility at Kahuta, but also including, among others, a research reactor with a small amount of U.S.-supplied HEU). [54] Pakistan's nuclear facilities are believed to be heavily guarded, though they probably are not equipped with state-of-the-art physical protection and material control and accounting technologies. [55]

Pakistani officials insist that they have taken a broad range of steps to beef up security and ensure that nothing comparable to A.Q. Khan's black-market nuclear exports can ever happen again, but have offered few specifics. Pakistan has reportedly established a security division headed by a two-star general under Pakistan's new Nuclear Command Authority; the division is reported to have 1,000 personnel (though this unit is to provide security against a broad range of threats, especially espionage, not just ensuring against theft of nuclear weapons and weapons-usable nuclear materials). [56] But Pakistan remains a society with a massive and deep-rooted problem of corruption, and this raises the same worrisome possibilities for short-circuiting security systems that exist in Russia; [57] while Pakistan reportedly now has extensive personnel screening and monitoring procedures in place, [58] it is unlikely that the nuclear enterprise can be entirely immune from the endemic problems facing the country.

Clearly, either state collapse or the rise of an extremist Islamic government in Pakistan — neither of which can by any means be ruled out — could pose severe dangers of nuclear assets becoming available to terrorists or hostile states.  Even in the current environment, however, both insider and outsider threats to Pakistan's stockpiles appear to be dangerously high — creating serious dangers despite the relatively modest size and relatively high levels of security of Pakistan's nuclear stockpiles.

Insider threats.  Recent events highlight the extraordinary danger that insiders in Pakistan's nuclear complex, motivated by money, sympathy to extreme Islamic causes, or both, might help terrorists get a bomb or bomb material from Pakistan's stockpiles.  First among these events are the extraordinary revelations concerning the global black-market nuclear network led by A.Q. Khan, the father of Pakistan's bomb, demonstrating that at least some nuclear insiders in Pakistan have been willing to sell practically anything to practically anyone — including designs and production manuals for uranium enrichment centrifuges, centrifuge components, operational centrifuges, and an apparently Chinese-origin nuclear bomb design. [59] The fact that the network was able to remove entire centrifuges from Pakistan's premier nuclear weapons material production facility and ship them off to other countries suggests either government approval or a truly extraordinary breakdown in security.  There is also the truly remarkable case in which Osama bin Laden and his deputy Ayman al-Zawahiri met at length with two senior Pakistani nuclear weapons experts with extreme Islamic views, and pressed them both about nuclear weapons and about others in Pakistan's program who might be willing to help.  Neither of these Pakistani scientists were ever tried or imprisoned, though it appears they remain under a loose form of house arrest.  Bin Laden may have been on the right track in asking for others who could help: by one estimate from a Pakistani physicist, some 10% of Pakistan's nuclear insiders are inclined to extreme Islamic views. [60] Finally, Pakistani investigations of the assassination attempts against President Musharraf in late 2003 suggest that they were carried out by military officers in league with al Qaeda operative Abu Faraj al-Libbi, raising disturbing possibilities for al Qaeda cooperation with the officers charged with guarding nuclear stockpiles. [61] In short, the danger that insiders might pass material or weapons to al Qaeda, or facilitate an outsider attack, appears to be very real.

Outsider threats. Similarly, the threat from a possible terrorist attack on a Pakistani nuclear weapon depot appears dangerously high.  Armed remnants of al Qaeda and of the Taliban continue to operate in the nearly lawless tribal zones on Pakistan's border with Afghanistan. As the July 2007 National Intelligence Estimate warned, Al Qaeda "has protected or regenerated key elements of its Homeland attack capability, including: a safehaven in the Pakistan Federally Administered Tribal Areas (FATA), operational lieutenants, and its top leadership." [62] Some combination of al Qaeda, Taliban, and Pakistani fighters was able to hold off thousands of Pakistani regular army troops for days at a time in a pitched battle in the tribal zones in early 2004. [63] If 41 heavily armed terrorists can strike without warning in the middle of Moscow, how many might appear at a Pakistani nuclear weapon storage site? Would the guards at the site be sufficient to hold them off — and would the guards choose to fight, to flee, or to cooperate?

Other Global Threats

The identification of these three categories as the highest priority threats is by no means intended to minimize the threats that exist elsewhere around the world.  There is probably no country where nuclear weapons and weapons-usable materials are located that does not have more to do to ensure that its nuclear stockpiles are secured and accounted for to a level sufficient to defeat demonstrated terrorist and criminal threats.  This is a global problem, which can only be solved through a global partnership for nuclear security.  Brief summaries of some of the other major stockpiles around the world follow below, beginning with the developing countries that possess nuclear weapons, continuing to the developed countries that possess nuclear weapons, and then considering the problem of civilian separated plutonium (civilian HEU having been discussed above).

China. While public information about China's approaches to nuclear security and accounting is sparse, China's nuclear security system is believed to be heavily dependent on "guards, guns, and gates," as the Soviet system was, with relatively little application of modern safeguards technologies. [64] China does not have a specific DBT defined in regulations, and systematic engineering approaches to assessing and correcting vulnerabilities are typically not applied. [65] As of October 2006, Chinese experts indicated that systems-engineering vulnerability assessments had not been performed at most sites, and were not required by Chinese regulations. [66] Chinese experts have expressed concern that improved protections against insider theft may be needed as China shifts toward a more market-oriented (and more corrupt) society. [67] Outside terrorist attack may someday also be an issue. China does have a continuing problem with terrorist groups, including groups based in China's Islamic minority, which the Chinese government believes are linked to al Qaeda.

The United States and China initiated a lab-to-lab cooperation program on technologies for securing and accounting for nuclear materials in the late 1990s, which ulti