Resolving the Ambiguity of Nuclear Weapons Costs

By Russell Rumbaugh and Nathan Cohn

For many observers, nuclear weapons top the list of areas in which savings should be found in the face of declining defense budgets. Indeed, the Department of Defense’s own strategic guidance acknowledges that “[i]t is possible that our deterrence goals can be achieved with a smaller nuclear force, which would reduce the number of nuclear weapons in our inventory as well as their role in U.S. national security strategy.”[1]

Fewer nuclear weapons presumably mean less spending, but no one can say how much less, in part because there is no definitive estimate of current nuclear weapons spending. Official estimates tend to be narrowly defined. Unofficial estimates capture more costs, but with less detail and authority. Without a definitive cost estimate, controversy over spending creates confusion, which clouds debate over U.S. nuclear posture and force structure.

During the past year, the Stimson Center conducted a study to help dispel that confusion.[2] The study found that most controversy stems from two specific issues. First, people disagree about what counts as nuclear weapons spending. Second, it is genuinely unclear what costs within the Defense Department support the operations and maintenance of nuclear weapons. Once these two issues are addressed, there is little disagreement about the cost of nuclear weapons. The study clarifies these two issues by comparing the various definitions of nuclear weapons spending, including those in official estimates, and using an inductive approach to estimate support costs within the Defense Department. Although the results still are not definitive, they clarify the debate and provide a framework for minimizing confusion about nuclear weapons spending. At the very least, this study should clarify that official estimates relying on a narrow definition of nuclear weapons understate the actual amounts the United States spends on nuclear weapons.

Defining the Spending

What counts as nuclear weapons spending has always been a thorny issue. The Obama administration’s best official estimate puts nuclear weapons spending at $214 billion over the next 10 years, or just more than $20 billion a year.[3] Yet, several independent studies say spending is as high as $55 billion a year.[4] These figures suggest a wide range of estimates for nuclear spending, stoking the controversy over how much nuclear weapons cost. The estimates differ largely because they count different things.

The responsibility for nuclear weapons costs was split more than 65 years ago and is still broken into two different parts run by two different agencies. Today, the National Nuclear Security Administration (NNSA), a semiautonomous unit of the Department of Energy, is in charge of the fissile material that makes a nuclear weapon nuclear.[5] The Defense Department is in charge of the delivery vehicles that would carry nuclear warheads to their target.[6] This split means no single agency in the federal government is responsible for all nuclear spending. There is no single line item for such spending in the annual budget justification documents. This is not to say that the agencies do not account for the spending. They do, but in ways that accord with their budgeting systems, which are not designed to answer a question such as how much the United States spends on nuclear weapons.

In the Energy Department, costs for nuclear weapons are displayed fairly straightforwardly. In fiscal year 2011, 63 percent of the department’s budget went to Atomic Energy Defense Activities, one of the formal subfunctions by which the U.S. government budget is displayed. The category further breaks down into two big groupings. The first, which accounts for $5.8 billion, or 36 percent, covers Defense Environmental Cleanup and other activities that clean up sites used to develop nuclear weapons. The other 64 percent, or $10.5 billion, goes to the NNSA. The NNSA itself runs four major programs that comprise that $10.5 billion budget: $2.3 billion is allotted to Defense Nuclear Nonproliferation activities, which aim to prevent the spread of nuclear material around the world; $1 billion for the Naval Reactors category, which covers the use of nuclear power for the propulsion of certain U.S. Navy ships; $400 million for the Office of the Administrator, covering NNSA central administrative functions; and $6.9 billion for Weapons Activities, to manage and maintain the stockpile of nuclear warheads.

There is a reasonable disagreement about which of these costs should count as nuclear weapons spending. Under a broad interpretation, all of these costs would qualify. After all, they all fall under Atomic Energy Defense Activities. The costs to clean up from nuclear weapons or prevent the spread of nuclear weapons would not be necessary if nuclear weapons never existed. Yet, under a narrow definition—one focused on the current U.S. nuclear weapons arsenal—these same costs may appear peripheral. A unilateral U.S. decision to end its nuclear weapons programs would have little impact on the policy choice of whether to continue to clean up previously used nuclear weapons sites or whether to try to prevent the spread of nuclear weapons elsewhere in the world.

Under a broad definition, all $16.3 billion of the expenditures under Atomic Energy Defense Activities would count as nuclear weapons spending. Under a very narrow definition of nuclear weapons spending, only the $6.9 billion of Weapons Activities would count. This difference of classification creates a swing of almost $10 billion in cost estimates.

That $10 billion range may seem broad, but it looks very narrow when compared to the possible range found in the Defense Department’s budget. The department had a base budget, not counting the funding for the wars in Afghanistan or Iraq, of $528 billion in fiscal year 2011, 20 times the size of the Energy Department’s budget and 32 times the size of Atomic Energy Defense Activities. Such a scale creates a huge pool of costs that could potentially be described as nuclear weapons spending. Still, as with the Energy Department, most disagreement over costs stems from definitions. Like the Energy Department, the Defense Department runs nonproliferation efforts—$1 billion worth. That figure would be included in a broad definition of nuclear weapons spending, but not a narrow one. The Pentagon also is responsible for the defense of the United States, including air defenses such as defenses focused on stopping enemy bombers and now ballistic missile defenses. In fiscal year 2011, the agency responsible for missile defense within the Defense Department had a budget of $8.4 billion. Some analyses would include these costs as part of nuclear weapons spending, and others would not.

Clarifying these definitional issues resolves most of the disagreement over the costs of nuclear weapons. The official estimate of $20 billion a year is based on a narrow definition. It can be doubled by using a much broader definition and including Energy Department programs beyond those in Weapons Activities and, in the Defense Department, programs such as missile defense. Both definitions are reasonable; they are just different definitions of what constitutes nuclear spending.

Once the official estimates are compared in a like manner to the unofficial estimates, most of the disagreement falls away. Nevertheless, there is a difference of roughly 27 percent between the official estimate and some unofficial estimates. This difference stems from the different treatment of the supporting costs of nuclear weapons.

Estimating Support Spending

Here then is real uncertainty about the costs of nuclear weapons. In the Energy Department, there is a core account of nuclear weapons costs, Weapons Activities. Other accounts could be considered nuclear weapons spending under some definitions—Defense Environmental Cleanup and Defense Nuclear Nonproliferation. Two accounts have some nuclear weapons spending even under a narrow definition—the Office of the Administrator and Naval Reactors. The administrator’s office provides central management of all NNSA programs, including weapons activities, but also including nonproliferation. Naval Reactors funds the development and maintenance of the Navy’s nuclear propulsion systems, including those on nuclear ballistic missile submarines, which carry U.S. nuclear weapons, but also including those on aircraft carriers, which do not carry nuclear weapons. It is difficult to determine what portions of these accounts support nuclear weapons. Still, these two accounts represent only $1.4 billion a year, limiting how much estimates can vary.

In contrast, estimates of spending in the Defense Department could vary across a wide range. It also has a core budgetary account of nuclear weapons costs, called Major Force Program 1 (MFP-1), which covers strategic forces, and it has some costs that could be considered nuclear weapons spending under certain definitions, such as nonproliferation and missile defense. Beyond those two accounts, however, the Defense Department’s budget includes another $500 billion, any part of which might support nuclear weapons.

To estimate what in this large pool of funding supports nuclear weapons, previous studies have used a deductive approach. They estimate the proportion that MFP-1 represents to other mission areas and then apply this proportion to the entire defense budget to determine support costs. For instance, if MFP-1 represents 10 percent of the costs of combat forces, then 10 percent of all support costs would be considered the nuclear weapons share of support costs. Such methodologies generate a reasonable but general estimate. This generality allows critics to charge that these costs are overstated, without having to contest specific costs, inflaming the confusion and thus controversy over nuclear weapons costs. To avoid this confusion, the Stimson study instead used an inductive approach to build a bottom-up list of specific costs.

In order to understand the study’s methodology and results, one must get a better sense of the defense budget. In the 1960s, the defense budget was divided into nine major force programs. Today, there are 11 major force programs, of which one is MFP-1. More than 40 percent, including war costs, of the defense budget is found in a single MFP, called general purpose forces, which are the combat forces that conduct conventional, or non-nuclear, operations.

In contrast, MFP-1 accounts for only 2 percent of the total defense budget. The rest of the budget is spread over the other nine MFPs. Most importantly, although the Defense Department still uses these MFPs to display the budget, they have not been used to manage the Defense Department or its budget for many years. Instead, managers in the Pentagon and overseers in Congress tend to use specific programs or appropriation titles to manage the allocation of funding. This lack of use means MFPs tend not to be completely accurate descriptions of the programs they are supposed to capture.

Nevertheless, MFP-1 does capture a great deal of Defense Department nuclear weapons spending. It includes the Air Force’s strategic bombers, the Air Force’s intercontinental ballistic missiles, and the Navy’s ballistic missile submarines. It includes the people and infrastructure that support those weapons systems, as well as the costs to operate and maintain them. It also includes a large part of the headquarters necessary for operations. MFP-1 underlies most estimates—official or unofficial—of spending on nuclear weapons. In fiscal year 2011, the budget amount for MFP-1 was $12 billion.

MFP-1, however, does not capture all of nuclear weapons spending under almost any definition. It does not include the cost of research, development, testing, and evaluation for programs dedicated to developing new or improved strategic delivery systems. For instance, the cost to research and develop the submarine designed to replace the Ohio-class ballistic missile submarines, known as the SSBN(X), as well as the cost to develop the next-generation bomber, is not included in MFP-1, but instead in the MFP that covers research and development funding. Yet, these development programs are direct costs of maintaining nuclear weapons.

MFP-1 also does not include many of the costs necessary to maintain the command and control of nuclear weapons, particularly the satellite-based communication systems and ballistic missile early-warning and detection systems. It also does not include many indirect but essential support costs, such as centralized supply and maintenance, and the strategic airlift necessary to transport nuclear weapons. Similarly, although it includes training for nuclear personnel and their pay, it does not include the costs of basic training, health care infrastructure, recruitment of nuclear personnel, or housing. Yet, these costs, like those of the Office of the Administrator for the NNSA, support not just nuclear weapons but other missions as well. For instance, the infrastructure that recruits and trains nuclear operators also recruits and trains operators of conventional capabilities. If the Defense Department’s accounting systems assigned these costs solely to the nuclear mission, it would then significantly underestimate the costs of the other missions. The Pentagon solves this problem reasonably by not assigning the costs to the nuclear or conventional missions. The costs are detailed in its annual budget justification documents and historical displays, but because they are not specifically assigned to nuclear weapons costs, they are not counted as the costs of nuclear weapons in official estimates.

To address these costs, the Stimson study made a theoretical distinction. It set out to determine the cost of the Pentagon capabilities necessary to maintain and operate nuclear weapons. Simply put, the study sought to assign a dollar figure to what would remain if the Defense Department’s only responsibility were to maintain and operate nuclear weapons. This approach is inherently broader than an approach that sought to determine the cost of support functions dedicated solely to nuclear weapons because many support functions also serve non-nuclear forces.

Using this broader standard of what counts as costs of nuclear weapons, the study looked at three categories outside MFP-1: command and control, other research and development, and other operational and support costs.

Command and control systems have three significant costs: procurement, research and development, and operations. Procurement costs are provided on a system-by-system basis, so the study listed each of the command and control systems relevant to nuclear weapons and summed their costs, finding $1.4 billion of procurement costs in fiscal year 2011. Research and development can be figured in the same way, producing a cost of $2 billion. Operating costs are not as simple because the costs for systems are aggregated in the budget displays; multiple systems are grouped in a single budget line. The study noted which systems were relevant to nuclear weapons and then proportionally calculated the nuclear weapons costs in each budget line. In all, the study found $2 billion in operating costs. For both operating and procurement costs, the study accepted the total cost of the system, even though most systems also support conventional operations. For example, a satellite may have additional capabilities to support nuclear operations. Because the satellite would not be functional without its entire system, however, the entire cost of the satellite is counted as part of the cost of nuclear weapons.

Research and development programs also are displayed in the budget justification documents by system, and as with command and control costs, the study summed up the costs for each relevant system. These costs are particularly important today because of the development programs to replace the bomber and submarine legs of the U.S. strategic triad. After years of fairly stable funding, these modernizations could change the scale of nuclear-related funding in the coming years, reaching possibly a quarter of the annual costs of nuclear weapons. In fiscal year 2011, the total research and development costs for nuclear weapons, not including command and control systems, were $1.5 billion. This figure will rise dramatically in the next 10 years as the follow-on systems advance. The Stimson study addresses these costs in another section on future costs, not addressed in this article.

Determining the other operational and support costs poses the greatest difficulty because they are not displayed in discrete ways as systems are under procurement or research and development. Yet, these costs are necessary to support nuclear weapons. The study looked at six specific components: other operational costs, training and recruiting costs, medical costs, family housing costs, centralized supply and maintenance costs, and centralized administrative costs. The  first category, other operational costs, includes the air refueling tankers that support the strategic bombers and a small amount that provides the nuclear-specific airlift to transport nuclear weapons. Tankers pose the single greatest challenge to estimating supporting costs because there is no public and official assignment of tanker responsibility between nuclear and conventional forces, and the current tanker fleet could be justified almost wholly by the nuclear or conventional mission. Moreover, the total costs of the tanker fleet are so large that the proportion counted toward the nuclear mission produces a wide swing in estimates of the total non-MFP-1 costs.

The study included only a narrow range of tanker costs: the costs to operate enough National Guard and Army Reserve tankers to provide two refuelings to all bombers designated for the nuclear mission. It does not include the modernization costs of the next-generation tanker currently being developed. Even this narrow definition produces about $500 million in costs. If all tanker costs were included, totally disregarding the conventional missions that currently dominate the operation of tankers today, the figure for fiscal year 2011 alone would be $6 billion.

Training and recruiting costs were calculated by determining the proportion of nuclear personnel to total military personnel and then applying that proportion to total training and recruiting costs. Health and housing costs were calculated in the same way. Centralized supply and maintenance costs were calculated using a base index to represent the proportion of infrastructure dedicated to nuclear weapons. Centralized administration was more difficult because war costs, which tend not to apply to nuclear costs, have skewed how centralized administration costs are displayed. In the budget displays, large sums of war costs are provided as contingency funds and captured in the central administration budget lines, creating spikes of funding from year to year. To compensate for this, the study averages projected costs for the next five years. In all, other operations and support costs come to $4 billion per year.

The study found $11 billion in other costs that support nuclear weapons within the Defense Department, besides the $12 billion of MFP-1 costs.

The study also addressed the two accounts within the NNSA that have some narrowly defined nuclear weapons spending and some funding that would be nuclear weapons spending only under a broader definition. The study proportionately calculated the costs of the Office of the Administrator based on the number of employees dedicated to the nonproliferation mission versus weapons activities.

Regarding Naval Reactors, much as with command and control systems, the study included the total cost even though some reactors support aircraft carriers, which do not have a nuclear mission. This decision was made because most of the infrastructure needed to produce and maintain reactors would be necessary regardless of the number of reactors. These two accounts include another $1.3 billion to add to the core NNSA Weapons Activities account’s $6.9 billion, producing a total NNSA cost of $8.2 billion.


By using a different methodology and relying on an inductive, bottom-up estimate, the Stimson study has validated the claim that there are costs that support nuclear weapons beyond those captured by official estimates. These costs are significant—74 percent more in costs than the core accounts alone. The study also helps bound the possible costs of nuclear weapons and thus helps demonstrate that there are not vast uncounted sums of nuclear weapons spending. Instead, most of the swing between various cost estimates stems from differences in definitions. It is perfectly reasonable to include missile defense and environmental cleanup as costs of nuclear weapons, but it is also reasonable to define the costs of nuclear weapons more narrowly. Still, the study has demonstrated that nuclear weapons do incur costs greater than the ones the official estimates describe, no matter how narrowly nuclear weapons spending is defined.

The Stimson study should ease some of the confusion surrounding the issue. Although it is not definitive, it does help distinguish cases in which estimates differ only because of different definitions from cases in which there is genuine uncertainty. In the latter cases, the study uses an approach that not only provides a reasonable estimate, but also makes its methodology clear so its conclusions can be debated analytically. Nuclear weapons are a grave matter, and all of their implications should be debated. This added clarity hopefully will improve the dialogue.

Russell Rumbaugh is director of the Stimson Center’s Budgeting for Foreign Affairs and Defense Program.

Nathan Cohn is a research assistant at the Stimson Center.


1. Department of Defense, “Sustaining U.S. Global Leadership: Priorities for 21st Century Defense,” January 2012.

2. Russell Rumbaugh and Nathan Cohn, “Resolving Ambiguity: Costing Nuclear Weapons,” Stimson Center, May 2012,

3. U.S. Department of Defense, “November 2010 Update to the National Defense Authorization Act of FY2010, Section 1251 Report, New START Treaty Framework and Nuclear Force Structure Plans,” 2010, This report refers to the 1251 report, but as it is a classified document, these references do not refer to the original report but to the aggregate of the reporting and unclassified summaries.

4. Stephen I. Schwartz, Atomic Audit: The Costs and Consequences of U.S. Nuclear Weapons Since 1940 (Washington, DC: Brookings Institution Press, 1998), Steven M. Kosiak, “Spending on U.S. Strategic Nuclear Forces: Plans and Options for the 21st Century,” Center for Strategic and Budgetary Assessments, 2006; Stephen I. Schwartz and Deepti Choubey, “Nuclear Security Spending: Assessing Costs, Examining Priorities,” Carnegie Endowment for International Peace, January 2009.

5. At the end of World War II, the Atomic Energy Commission was created to house civilian control of military nuclear programs. In 1974 the commission was dissolved, and oversight responsibility for the national nuclear laboratories was transferred to the newly created Energy Research and Development Administration, which later merged with other federal entities to become the Department of Energy. In 1999, in the aftermath of allegations that lax oversight and security had culminated in the loss of nuclear secrets to China, the National Nuclear Security Administration was created and assumed responsibility for the national security responsibilities of the Energy Department.

6. Other agencies also have programs that could be included as nuclear weapons spending, but this article will focus only on the departments of Defense and Energy.

New layer...