What US importers actually paid
One number, and it is not the number in the headlines. This is duty assessed on goods entered for consumption, divided by the value that was actually subject to duty — counted by US Census, not estimated by us, every month since January 2017.
- On the dutiable part
- 15.66% in June 2026
- On everything imported
- 6.81% — the same duty over all goods entered for consumption, most of which pay nothing
- The bill
- $21.29bn of duty assessed on $135.98bn of dutiable value, out of $312.48bn entered
- Months on the record
- 114 — January 2017 to June 2026, no gaps
- Underneath it
- 19,850,923 measured lines — code × origin × month
Duty divided by the taxed portion is not duty divided by everything shipped. The first is roughly double the second and the two get quoted as though they were one figure. Every number on this site names which it is.
Assessed, not collected
Census computes duty from the rates at entry. It excludes anti-dumping and countervailing duties and it excludes fees, and no cash has moved when the number is struck. Treasury cash is a different book.
Year by year
| year | months | entered for consumption | dutiable value | duty assessed | on the dutiable part | on everything |
|---|---|---|---|---|---|---|
| 2026 | 6 | $1.73tn | $722.49bn | $129.86bn | 17.97% | 7.52% |
| 2025 | 12 | $3.39tn | $1.33tn | $259.37bn | 19.56% | 7.65% |
| 2024 | 12 | $3.25tn | $1.01tn | $76.63bn | 7.59% | 2.36% |
| 2023 | 12 | $3.07tn | $975.05bn | $72.15bn | 7.40% | 2.35% |
| 2022 | 12 | $3.22tn | $1.09tn | $89.76bn | 8.22% | 2.78% |
| 2021 | 12 | $2.82tn | $940.04bn | $83.31bn | 8.86% | 2.96% |
| 2020 | 12 | $2.33tn | $725.23bn | $64.36bn | 8.88% | 2.76% |
| 2019 | 12 | $2.50tn | $844.92bn | $66.12bn | 7.83% | 2.65% |
| 2018 | 12 | $2.55tn | $834.06bn | $46.42bn | 5.57% | 1.82% |
| 2017 | 12 | $2.33tn | $706.30bn | $32.94bn | 4.66% | 1.42% |
2026 is 6 months, not twelve. It is not comparable to a full year and is not presented as one.
The last two years, month by month
| month | entered for consumption | dutiable value | duty assessed | on the dutiable part |
|---|---|---|---|---|
| 2026-06 | $312.48bn | $135.98bn | $21.29bn | 15.66% |
| 2026-05 | $305.19bn | $129.11bn | $21.03bn | 16.29% |
| 2026-04 | $299.71bn | $124.53bn | $20.20bn | 16.22% |
| 2026-03 | $299.17bn | $123.93bn | $20.49bn | 16.53% |
| 2026-02 | $250.56bn | $101.44bn | $21.24bn | 20.94% |
| 2026-01 | $259.41bn | $107.50bn | $25.61bn | 23.82% |
| 2025-12 | $277.69bn | $113.76bn | $25.94bn | 22.80% |
| 2025-11 | $257.38bn | $105.34bn | $25.20bn | 23.92% |
| 2025-10 | $274.47bn | $117.43bn | $29.74bn | 25.33% |
| 2025-09 | $275.54bn | $117.25bn | $29.03bn | 24.75% |
| 2025-08 | $256.33bn | $118.19bn | $26.89bn | 22.75% |
| 2025-07 | $289.83bn | $133.48bn | $27.55bn | 20.64% |
| 2025-06 | $259.20bn | $127.29bn | $24.07bn | 18.91% |
| 2025-05 | $272.74bn | $123.25bn | $24.71bn | 20.05% |
| 2025-04 | $273.42bn | $105.85bn | $19.64bn | 18.55% |
| 2025-03 | $340.67bn | $93.64bn | $12.67bn | 13.53% |
| 2025-02 | $292.03bn | $82.38bn | $6.81bn | 8.26% |
| 2025-01 | $321.91bn | $88.39bn | $7.13bn | 8.07% |
| 2024-12 | $281.59bn | $84.99bn | $6.58bn | 7.74% |
| 2024-11 | $271.43bn | $83.47bn | $6.38bn | 7.65% |
| 2024-10 | $291.25bn | $86.81bn | $6.85bn | 7.89% |
| 2024-09 | $283.60bn | $85.60bn | $6.82bn | 7.97% |
| 2024-08 | $275.69bn | $87.60bn | $6.93bn | 7.91% |
| 2024-07 | $284.67bn | $91.86bn | $7.07bn | 7.70% |
Take the file
Where this comes from, and how to check it without us
I_COMMODITY, GEN_VAL_MO, CON_VAL_MO, DUT_VAL_MO, CAL_DUT_MO
rate on the dutiable part = CAL_DUT_MO ÷ DUT_VAL_MO
rate on everything = CAL_DUT_MO ÷ CON_VAL_MO
The Census key is free and nothing above needs anything from us. If our arithmetic is wrong you can establish that without our cooperation, which is the only kind of check worth publishing.
Every month here has been checked against the Census API’s own world-total row for the same month: 114 of 114, worst deviation 0.000%. That proves the pull is complete. It proves nothing external — it cannot fail unless we dropped rows. Separately, duty assessed at entry against US Treasury gross customs receipts sits at 0.85–0.95 in every year from 2017 to 2026, and the gap between those two books is explained rather than residual. The full arithmetic.
Send one entry
A national average is the wrong resolution for your work. The same measurement, applied to one of your own entries, is a page of lines rather than a headline.
Free. No account. We do not file with CBP and take no fee on anything you recover. Your file is read on our own server and is never sold or shared.
one code, one origin · how a line is priced · what this does not do