Every ranking of America's trade deficit counts goods and drops the $327.7bn the country wins back

Twelve countries, twelve deficits, and a total that does not hold them. Google's AI Overview for the phrase people actually type lists America's largest bilateral goods deficits, from China at $202.1bn down to Malaysia at $30.8bn, and reports the overall deficit as nearly $900bn. Add the twelve figures up. They come to $1,200.1bn.

Nothing there is a mistake, exactly. The two halves measure different things, and the reader is not told which is which. The country list counts goods. The overall figure counts goods and services together, and the services line is a surplus large enough to change who is on the list and in what order.

The June 2026 release from the Census Bureau and the Bureau of Economic Analysis publishes both, in adjacent exhibits, on the same morning. Almost nobody reads the second one.

What is the actual US trade deficit?

The United States ran a goods and services deficit of $931.96bn in 2025. That was a deficit of $1,259.62bn on goods and a surplus of $327.66bn on services. The combined figure is 3.03 percent of the $30.76tn economy. The goods figure alone is 4.09 percent, and quoting it as the deficit overstates the gap by a third.

Both numbers are official and both are correct. They are not interchangeable, and the difference between them is $327.66bn, which is roughly the annual output of Denmark.

The services surplus is not a rounding item. It offsets 26.0 percent of the goods deficit. It has offset something like a quarter of it for most of this century, and it is the single largest reason a bilateral goods table misleads.

One more piece of context before the countries. The United States has run a combined deficit every year since 1976, fifty years in a row. The last surplus was 1975, at $12.4bn. Services were themselves in deficit until 1971, when they turned positive by $959m and stayed there.

Why do the country rankings count only goods?

Because goods and services are collected by different agencies, in different ways, on different schedules. Goods data come from the customs declarations that U.S. Customs and Border Protection collects on every shipment, which yields a country of origin or destination for each one. Services are surveyed by BEA and published quarterly, not monthly.

That gap in timeliness is the whole explanation, and it hardens into a habit. Monthly goods figures by country are available five weeks after the month ends. Country detail on services arrives quarterly and later. A ranking that wants to be current uses the goods file, and once the goods file is the source, the services column is simply absent from the conversation.

There is a second reason, and it is less forgivable. The goods country table and the goods and services country table are published in the same document, and the second one is number 20 out of twenty-two exhibits.

Which countries does the United States run its largest deficits with?

Mexico, at $190.12bn in 2025 on the combined measure, then Vietnam at $177.11bn and China at $167.61bn. On goods alone the order is Mexico, China, Vietnam. Services move China from second to third, cut Ireland from fifth to eighth, and shrink Switzerland by three quarters.

Here are the fifteen largest deficits, ordered two ways. All figures are 2025, balance of payments basis, in millions of dollars, as published by BEA.

Eight of the fifteen sit in a different position depending on which column you read. That is not a technicality about a rounding convention. It is a different answer to the question the reader asked.

The surplus side never appears in a deficit ranking at all, which is its own kind of editing. On the combined measure the United States ran surpluses of $81.02bn with the Netherlands, $45.46bn with the United Kingdom, $41.82bn with Brazil, $33.10bn with Singapore and $30.26bn with Hong Kong.

Where do services change the answer most?

Ireland, by a distance. The goods deficit with Ireland was $114.47bn in 2025 and the services surplus was $66.29bn, so the combined deficit was $48.18bn, smaller by 57.9 percent. Switzerland shrinks by 74.7 percent, Canada by 50.3 percent, and China by 17.0 percent.

Ireland is the case that breaks the ranking, so it is worth staying with. Fifth-largest goods deficit in the country, eighth once services count. And the direction of travel is sharper than the annual figure suggests: in the fourth quarter of 2025 the combined balance with Ireland was a surplus of $8.76bn, and in the first quarter of 2026 a surplus of $9.40bn, because the quarterly goods deficit collapsed to $8.14bn while the services surplus held at $17.54bn.

That is not a small revision to a story. On the measure that counts everything the United States buys and sells, Ireland stopped being a deficit partner two quarters ago.

Canada is the other one worth knowing. The combined balance with Canada was a surplus in 2013, 2015, 2016, 2017, 2018 and 2020, in every one of which the goods balance was a deficit. Six years in which the answer to "does the United States run a trade deficit with Canada" was yes and no at the same time, depending on the column.

The European Union follows the same rule at scale. Goods deficit of $222.99bn, services surplus of $107.28bn, combined deficit of $115.71bn. Counting services cuts the number by 48.1 percent.

Are there countries where services make it worse?

Yes, five of them, and they are the reason this is a correction rather than an argument. The United States runs services deficits with Italy, India, Germany, France and Israel. The combined deficit with each is therefore larger than the goods figure: 17.6 percent larger with Italy, 10.5 percent with France, 8.1 percent with India, 5.0 percent with Israel and 4.3 percent with Germany.

India is the instructive one. The goods deficit was $58.16bn in 2025 and the services balance was a deficit of $4.73bn, so the true figure is $62.89bn. India moves up the table, from eighth by goods to sixth by goods and services, passing Japan and Ireland.

Anyone using the services column to argue the deficit is smaller than reported should note that it does not do that everywhere. It does it in most places and the reverse in a few, which is what an honest adjustment looks like.

What is in the $327.7bn services surplus?

Three categories carry it: financial services at $141.94bn, charges for the use of intellectual property at $133.86bn, and other business services at $106.96bn. Together those come to $382.75bn, which is more than the whole surplus, because two large categories run the other way.

Transport is a deficit of $22.19bn. Insurance is a deficit of $102.32bn, and it widened by $27.16bn in a single year as insurance imports rose from $105.32bn to $136.51bn.

The travel line deserves a note, because it moved and almost nobody reported it. The travel surplus fell from $34.19bn in 2024 to $21.06bn in 2025, down 38.4 percent, as travel imports, meaning American spending abroad, rose $14.41bn while travel exports rose $1.28bn.

Put those together and 2025 produced an odd result. Services exports rose $81.22bn, or 7.0 percent. The services surplus rose $345m. Almost the entire gain in exports was absorbed by insurance and travel moving the other way, and a headline reporting record services exports would have been true and would have told the reader nothing about the balance.

Do the country numbers add up to the national total?

One set does and one set does not, and the release says so in its own explanatory notes. The BEA country table on a balance of payments basis sums to $931,963m against a published total of $931,960m, a difference of $3m in $932bn. The Census-basis country table does not sum to the national total by construction.

The note is worth reading in the Bureau's own words. Of its seasonally adjusted country exhibit it says the data "will not sum to the seasonally adjusted commodity-based totals because the seasonally adjusted country and world area data and the commodity-based totals are derived from different aggregations of the export and import data and from different seasonal adjustment models", and it adds that users "should use caution drawing comparisons between the two sets of seasonally adjusted series."

There are also real definitional gaps between the two bases, and Canada is the clearest. On a Census basis the 2025 goods deficit with Canada was $48.30bn. On a balance of payments basis it was $55.05bn. The largest single reason is that BEA adds inland freight charges for goods moving from Canada and Mexico to the US border, because customs values those shipments at the point of origin rather than at the frontier, unlike goods from anywhere else. Mexico moves the same way, from $197.03bn to $203.21bn.

So the country table that adds up is the one nobody quotes, and it is one exhibit further down the same file.

What has changed in 2026?

A great deal, and quickly. Through June the goods and services deficit was $371.23bn, down $189.31bn or 33.8 percent from the same six months of 2025. Exports rose $198.3bn, or 11.7 percent. Imports rose $9.0bn, or 0.4 percent.

The country order has changed with it. On the Bureau's own unadjusted goods ranking for the first half of 2026, the largest deficit is with Vietnam at $114.0bn, then Taiwan at $107.2bn, Mexico at $102.6bn and China at $73.9bn. Thailand is fifth at $54.2bn, ahead of South Korea, Germany, Canada, India and Japan.

China is fourth. Every ranking in general circulation still has it first, and that was true of the full year 2025 on goods and was not true of 2025 on goods and services, where Vietnam had already passed it.

One limit, stated where it bites rather than at the end. The combined goods and services country data are quarterly and run only through the first quarter of 2026, so the 2026 reordering above is a goods-only picture. The services correction that moved eight of fifteen places in 2025 has not yet been applied to this year, and when it is, the order will move again.

Why does the Bureau's own historical file disagree with its current release?

Because it has not been regenerated since February. The Census Bureau publishes a long series titled "U.S. Trade in Goods and Services, 1960 through 2025" as a one-page PDF. It gives the 2025 deficit as $901.47bn. The current release gives $931.96bn. The gap is $30.49bn.

The file carries a creation date of 18 February 2026 and the server reports it unchanged since 19 February. The current exhibit was published on 4 August 2026 and incorporates the annual revisions the release describes, and BEA's own 1960-forward file, updated the same day, matches the current exhibit to the dollar. Two current government files agree; the February one is the odd one out.

The difference is not confined to the total. The February file puts the 2025 services surplus at $339.47bn against the current $327.66bn, and it shows that surplus rising $27.6bn in 2025. On the current data the surplus was flat, up $345m. Those are opposite descriptions of the same year, both published by the same agency, both reachable today, and only one of them is current.

Older years moved too. The February file gives 2023 as $774.21bn and 2019 as $559.27bn, against $749.59bn and $535.41bn now.

What this changes

If you are reading a bilateral number, check which one it is before you use it. A goods figure and a goods and services figure differ by a quarter at the national level and by more than half for individual partners, and both are published as "the trade deficit".

If you are comparing the deficit to the economy, use the combined figure. It was 3.03 percent of GDP in 2025, against 3.02 percent in 2024. The goods-only ratio of 4.09 percent is a real number describing a real thing, and it is not the trade deficit.

And if a table you are reading ranks countries by deficit and shows no surpluses, remember what has been left out. The Netherlands, the United Kingdom, Brazil, Singapore, Hong Kong, Australia, Saudi Arabia and Belgium are all net buyers from the United States on the combined measure, and Ireland has been for two quarters. A ranking of deficits is a ranking of deficits. It is not a ranking of trading relationships, and it was never designed to be read as one.

The parallel is exact with the ownership of the federal debt, where the monthly Treasury table that everyone quotes is a custodial ledger rather than a register of owners, and reading it as the latter produces confident wrong answers. We wrote that up in how much US debt foreign countries hold. The pattern is the same: the number is fine, the label on it is not.

FAQ

Which country has the largest trade deficit with the United States? Mexico, on both measures, at $190.12bn on goods and services in 2025 and $203.21bn on goods alone. Vietnam is second on the combined measure at $177.11bn and China third at $167.61bn. On goods alone China is second and Vietnam third.

Is the US trade deficit with China still the biggest? No. It was the largest goods deficit in 2024 and it was second to Mexico in 2025, third once services are counted, and fourth in the first half of 2026 on the Bureau's unadjusted goods ranking, behind Vietnam, Taiwan and Mexico. The goods deficit with China peaked at $417.27bn in 2018 and was $202.04bn in 2025.

Why is the trade deficit reported as $931.96bn in one place and $1.26tn in another? The first counts goods and services together and the second counts goods only. The United States runs a services surplus of $327.66bn, and whether it is included is the entire difference. Both figures are published in the same monthly release.

Does the United States run a trade surplus with any large country? Yes. In 2025 the combined surplus was $81.02bn with the Netherlands, $45.46bn with the United Kingdom, $41.82bn with Brazil, $33.10bn with Singapore, $30.26bn with Hong Kong, $19.35bn with Australia, $13.66bn with Saudi Arabia and $9.14bn with Belgium.

How large is the trade deficit relative to the economy? The goods and services deficit was 3.03 percent of current-dollar GDP in 2025, against 3.02 percent in 2024. Through the first half of 2026 the deficit is running 33.8 percent below the same period of 2025.

Where can I get the underlying data? Exhibits 1, 19, 20, 20a and 20b of the monthly FT-900 release, and BEA's geographic time series file, which carries country balances on goods, services and both from 1999 forward. The two tables built for this piece are linked above as CSV.