How to Read a Munitions Bill: What Pentagon and Budget Office Numbers Actually Measure

Read almost any coverage of a munitions shortage and you will find two large numbers attached to the same conflict, which appear to contradict each other. One comes from an inspector general inside the Pentagon. The other comes from the Congressional Budget Office outside it. Neither is a mistake, and the gap between them is not evidence that somebody is lying.

The gap exists because the two documents count different objects. One is counting what was spent. The other is counting what it will cost to replace. Learn to tell those apart and most of the confusion in this field dissolves.

This guide works through the distinctions that matter, using real documents from the current Iran conflict, which happened to produce the most detailed public munitions accounting of any recent war. Worked examples are drawn from the Pentagon inspector general's report of 15 September 2026 and the Congressional Budget Office's analysis published the same day.

Identify who is speaking before you read the number

Two institutions dominate this material and they have deliberately different vantage points.

A Department of Defense inspector general report is an internal product with statutory standing. Inspector generals are appointed to produce reports to Congress precisely so that findings can be made that the department's own spokespeople would not make. The inspector general can ask the spending organisation for its files, and generally does. Its blind spot is that it inherits the department's own accounting, and its reports stop at a fixed cut-off date because a report is a snapshot.

A Congressional Budget Office analysis is an external product from a nonpartisan legislative office. Its authority comes from being able to say that an estimate is unreliable precisely because the subject refused to provide data. Its blind spot is that it must reconstruct, sometimes from public databases and reported figures, and it says so. Its reports extend the period covered because there is no internal reporting deadline.

The useful habit is to identify which of these you are reading before you read anything else, because it determines which question the number is answering.

The words that do the work

Most confusion in this field comes from six terms being used loosely. Here is what each one measures, and what it does not.

Term	What it measures	What it does not measure
Cumulative obligations	Money legally committed through contracts and binding actions	Money spent, or money approved
Expended munitions	Acquisition value of munitions actually consumed	What replacing them will cost
Equipment losses	Damaged and destroyed equipment, broadly read	Which items must be bought new
Equipment lost in battle	Equipment destroyed	Equipment that was damaged and repaired
Replacement cost	Forward estimate of the cost to buy back what was expended	What was actually spent
Unit cost	Acquisition cost of one missile	Launcher, radar, ship, crew, annual sustainment

The last row matters more than it looks. A published unit cost is the price of the missile alone. It excludes the vehicle that carries it, the sensors that find the target, the people who operate both, and the next round of procurement that keeps the system alive. Two documents quoting wildly different unit prices for the same missile are usually not contradicting each other; one is quoting procurement cost and the other is quoting a broader programme or unit cost, and only the first is comparable.

There is a related set of three terms that describe production and are constantly confused.

A buy rate is the number of missiles a budget requests or authorises in a year. It is an intention expressed in a document.

A production rate is what a factory is actually building in a year. It is an operating fact, and it is usually behind the buy rate because it takes years to move.

A delivery is what arrived last year. Deliveries lag production, because missiles built in one year are frequently not accepted until the next.

A sentence that says production has been raised to 2,000 a year and a sentence that says 2,000 were delivered are not saying the same thing, and the distance between them is where procurement reporting usually goes wrong.

Expenditure and replacement are different numbers

Here is the single most important distinction in the field, and it is worth holding onto even when it costs you a punchy headline.

Expenditure is retrospective. It is what the department paid for the rounds it fired.

Replacement cost is prospective. It is what those rounds will cost to buy again, valued at prices that do not exist yet, for a mix that someone must still decide.

A dollar of expenditure therefore does not buy a dollar of replacement, and this is not a technicality. Munitions bought in peacetime at contracted prices cannot be replaced on the same terms after the order books have been rebuilt and the labour market has moved. One published example of this dynamic: after Red Sea operations that were a fraction of the Iran war by any measure, the US Navy required more than $2 billion to replenish $1 billion of munitions expended.

When you see a figure labelled as the cost of replacing munitions, do not report it as the cost of the fighting. When you see a figure labelled as munitions expended, do not treat it as the size of the bill that is coming.

The interval test

Two totals are not comparable until you have found both cut-off dates, because the cut-off is part of the number.

Worked example. A Pentagon inspector general report gives $33.4 billion for combat operations through 29 June 2026. A Congressional Budget Office analysis gives $38.1 billion through 1 August 2026. The headlines differ by $4.7 billion and the intervals differ by about five weeks.

The first calculation a reader should make is the implied run rate, which here is roughly $4 billion a month.

The second calculation is to compare that implied rate against each source's own forward projection for the same period. The budget office separately estimated roughly $2 billion a month at the level of violence seen in May and June, and about $3 billion a month at July's intensity.

The implied rate is higher than either. That does not mean somebody is wrong. It more likely means the two documents cover different perimeters, because an inspector general report on a single named operation is narrower than a budget office estimate of combat operations against a country generally. The correct conclusion is not that the numbers are suspect. It is that they are not measuring the same thing, which is itself the finding.

A useful rule: never subtract two defence cost totals whose cut-off dates you have not written down next to each other.

When the count is withheld

The most direct way to check a claim about stockpile depletion would be to read the stockpile. You cannot, because the Pentagon does not disclose how many missile defence interceptors it holds, and has not been required to.

This is the moment at which a reader has to decide how much confidence to place in a depletion estimate, and it is worth being clear about what the standard workaround does and does not establish.

The Congressional Budget Office's approach was to compare reported expenditure against the total number of missiles the department has ever purchased, and infer what share of the stockpile that spending implies.

Read the logic carefully. Total purchases is a ceiling on inventory, not a measurement of it. Between the two sit munitions used in training, components pulled for inspection, and rounds consumed in exercises, none of which are in the expenditure figure for a conflict but all of which are absent from usable stock. The inference is therefore closer to an upper bound on the shortfall than a measurement of it, and a report that says so explicitly is being careful.

That is why a careful report publishes a range rather than a point estimate. The published finding, that between one-half and two-thirds of the combined Patriot, THAAD, SM-3 and SM-6 inventory has been used since June 2025, spans a serious problem and a catastrophic one. Any account that picks a point inside that band and presents it as the finding has misread the document.

Two further limits are worth carrying. The figure pools four interceptor systems together, so a reader cannot tell from the report whether most THAADs remain while most SM-6s are gone, which are very different worlds. And a deficit that predates a conflict is not caused by it, which is a separate check you must run rather than assume.

The exchange rate test

If you take only one calculation from this guide, take this one, because it requires no inventory data and survives every dispute about stockpile counts.

Divide the cost of the defensive missile by the cost of the target it intercepted.

Worked example: a Patriot interceptor costing roughly $4 million used against a drone costing roughly $35,000 produces a ratio of about 114 to 1. Reporting from the same conflict described coalition batteries firing eight interceptors against a single drone in some engagements, which is worse.

Why this ratio is so robust is that both inputs are independently verifiable. One comes from published unit costs. The other comes from the observed value of what was targeted. Neither depends on a classified inventory.

And it explains the industrial pressure better than any total does. A defender who loses 114 dollars of ammunition for every dollar of attack cannot sustain that posture for long, regardless of how much money the treasury authorises, because the constraint is how fast the magazine can be refilled. It also points directly at the only real remedy, which is layering cheaper systems beneath the expensive one.

The lead time test

A budget line is not a capability, and a contract is not a delivery.

The distinction is easy to state and easy to forget under deadline pressure. Production lead times published for the Patriot are 24 months for the missile and 30 months for its solid rocket motor, reflecting curing times for the motor and multi-year qualification of new suppliers.

This produces results that look absurd and are entirely normal. An emergency contract announced in April 2026 was not expected to deliver interceptors before mid-2028. The emergency purchase will arrive roughly two years after the emergency it was bought to answer.

The correct reading of any procurement announcement is therefore a three-part question. How many units, at what unit price, and by when. An announcement that gives you a dollar figure and a press release but no delivery schedule has told you about intent.

A fourth question is worth adding in a period of continuing appropriations: is there an appropriation behind it. A multiyear production commitment signed against a continuing resolution is a promise contingent on a Congress that has not yet voted, which is a real and underreported category of military procurement news.

A ten minute verification routine

When a munitions figure reaches you, in any medium, run this.

Find the primary document. Pentagon inspector general reports and Congressional Budget Office analyses are published as free PDFs on their own websites, and a search on the agency name plus the operation or programme will usually find it within a minute.

Read the cover or the first paragraph for the cut-off date and the scope. Write both down. Almost half of all apparent contradictions dissolve at this step.

Find the exact phrase the figure came from. A number in a news story should be traceable to a sentence in the document. If the document says "probably used between one-half and two-thirds", the story does not get to say "two thirds".

Check the verb. Expended, replaced, lost, damaged, requested, authorised, obligated and delivered are seven different verbs and only some of them describe money that has moved.

Check the exclusions. If the document excludes costs for personnel, for base repairs, or for other agencies, then the figure is a floor, and describing it without the exclusions overstates what is known.

Do the interval arithmetic if you have two figures. If the implied run rate exceeds what either source projected, look for a scope difference before concluding that anything is concealed.

Six figures that should make you stop

A total with no cut-off date anywhere near it. Ask when the estimate stops.

A replacement cost quoted as a cost. It is a forward valuation of a bill that has not arrived.

A unit cost with no fiscal year and no qualifier. Procurement-only cost and full programme cost differ substantially, and so do different years' appropriations.

An inventory number attributed to a briefing rather than a document. If the count is not published, a figure circulating in press reporting is usually a derived estimate, and it should be described as one.

A production rate that is actually a target. Framework agreements and multiyear procurement announcements state intentions with future dates attached, and those dates are routinely reported as though they were deliveries.

A cost that excludes the wounded, the dead, veterans' care and base repairs, presented as the cost of a war. Every defence ledger in this field excludes those, and every one of them is a real cost borne somewhere other than the budget line.

A short glossary

Buy rate. Missiles requested or authorised per year in a budget. An intention.

Cumulative obligation. Money legally committed through contracts. Not the same as money spent.

Delivery. Missiles that arrived. Lags production by a year or more.

Expended munitions. Munitions actually consumed, valued at acquisition cost.

Industrial base. The network of suppliers below the prime contractor, which is where actual output is determined.

Lead time. Months or years from a decision to build something to the moment it is delivered and accepted.

Multiyear procurement. A contract spanning several years, designed to give industry the stable demand signal it needs to invest in capacity.

Production rate. What a factory builds in a year. An operating fact.

Replacement cost. Forward estimate of what it will cost to buy back what was expended.

Sub-tier supplier. A firm beneath the prime contractor. Modern missile output moves at the speed of its weakest indispensable supplier, which is nearly always here.

Unit cost. The price of one missile, excluding everything that carries, finds or sustains it.

Where the documents are

Inspector general reports are published on the relevant inspector general's own website and are searchable by fiscal year and operation name. Congressional Budget Office analyses are published free on cbo.gov, and each one states its requester, usually a member of Congress, which tells you what question it was written to answer.

Both types are worth reading in the original even when a summary is available, because the caveats live in a footnote or a limitations paragraph that no summary reproduces. In the documents discussed here, the single most consequential sentence about the limits of the analysis is a short paragraph explaining that the department declined to respond to the agency's information requests, and that the estimate therefore relies on government databases and public reporting and is subject to considerable uncertainty.

That sentence is the most trustworthy part of the whole exercise, and it is the one most often left out of coverage.

For how these documents sit inside the wider budget process, including where a funding request becomes an appropriation, see our guide to military construction budgeting and the step-by-step walkthrough of a Pentagon budget volume in searching the C-1. The numbers discussed at length here are analysed in our report on the munitions and interceptor arithmetic.