Poor-condition bridges kept declining after the 2021 infrastructure law, but at roughly half the pace of the decade before it

Summary

By the one national metric FHWA actually publishes for every bridge every year — the share rated structurally Poor — America's bridges have kept improving since the Infrastructure Investment and Jobs Act (IIJA) was signed in November 2021, but the pace of improvement slowed, it did not accelerate. The poor-bridge count fell from 58,160 in 2011 to 43,586 in 2021 (a 2.8%/year decline over that decade); since IIJA it has fallen only from 43,586 to 41,685 through 2025 (a 1.1%/year decline) — roughly half the pre-law rate. The poor share of the inventory dropped 2.6 percentage points (9.61% → 7.03%) in the decade before the law, versus 0.4 points (7.03% → 6.68%) in the not-quite-4 years since. This holds whether you compare the full pre-law decade or a length-matched 4-year window (2017-2021 fell 2.2%/year vs. 2021-2025's 1.1%/year). The deeper story: the national fleet is aging into Fair condition faster than Poor bridges are being fixed — Fair share has risen every single year in this series, from 43.1% (2011) to 49.6% (2025), while Good share has fallen from 47.3% to 43.7%. So "improvement" here means the worst bridges keep shrinking as a share of the total, slowly — not that the system as a whole is getting newer.

Why 'poor share' and not some other measure

FHWA's National Bridge Inventory (NBI) is the only complete, annual, bridge-by-bridge condition census in the US, and its own condition_ratings roll up into a single field — bridge_condition — coded Good/Fair/Poor based on the lowest deck/superstructure/substructure/culvert rating. That field, and the state-by-state 'Bridge Condition by Highway System' tables FHWA itself publishes from it, are the standard the field (and IIJA's own performance targets) use to track bridge health, so this is the natural operationalization of 'how much have bridges improved.'

The AskAmerica warehouse (transport.bridges) carries this table for 2017-2025 with bridge_condition populated from 2019 onward (2017-2018 rows are present but the condition rollup was not computed in-warehouse for those two years). To reach the full decade before IIJA and to fill 2017-2018, this analysis reads FHWA's own published state-by-state national totals tables directly (fhwa.dot.gov/bridge/nbi/no10/condition{YY}.cfm) for 2011, 2017, 2021 and 2025. As a validity check, the warehouse's own 2021-2025 counts were queried directly and matched FHWA's published totals for 2021 exactly (43,586 poor bridges both ways) — the two sources agree.

The numbers, decade-before vs. since-the-law

Both the full-decade baseline and a window matched in length to the post-law period (per this corpus's own before/after guidance, an asymmetric window can manufacture an artificial finding, so both are reported) show the same thing: the rate roughly halved after the law took effect, it did not improve. By total deck area rather than bridge count, the poor share fell from 8.2% (2011) to 5.1% (2021) to 4.9% (2025) — same deceleration pattern.

The overall bridge inventory also grew slightly across both periods (605,103 → 619,622 → 624,193), so a small part of the 'poor count' improvement in both eras is diluted by new bridges entering the population as Good, not existing Poor bridges being fixed — this analysis reports both raw counts and shares for that reason.

Why the deceleration despite record funding — an obligation-lag explanation, not a data artifact

The IIJA created the Bridge Formula Program: $26.5 billion over five years (FY2022-2026) for roughly 15,000 bridge repair/replacement/rehabilitation projects, plus a further $12.5 billion discretionary Bridge Investment Program. But federal condition data will only show the effect of that money once projects are physically completed and re-inspected, and GAO's reporting on IIJA implementation found obligation running well behind appropriation: as of April 2024 only about $7 billion of the roughly $16 billion in bridge-formula funds available to that point had been obligated to specific projects, and states have up to four years to obligate formula funds. As of April 2025, DOT overall had obligated 59% of available IIJA funding. That lag is consistent with what the NBI data shows through 2025: money is moving, but the physical-condition payoff for a multi-year construction pipeline mostly falls after this report's data window (2025), which is why the observed deceleration in poor-bridge reduction should be read as 'too early to show up yet,' not as evidence the program failed.

Caveats

• The AskAmerica connector's own transport.bridges table only carries NBI data for 2017-2025 with a computed bridge_condition field from 2019 forward; the 2011 and 2017 figures in this report come directly from FHWA's own published national tables (web-fetched from fhwa.dot.gov), not the warehouse, and are cited as such below.
• 2018 is omitted from the trend line because the connector's copy has a null condition rollup for that year and a targeted FHWA fetch for it was not performed; this does not affect either the decade or 4-year annualized-rate comparisons, which use only years with confirmed figures.
• The Bridge Formula Program funding/obligation figures are general programmatic totals (not bridge-condition-specific spend-to-completion tracking) and are meant to explain the mechanism of the lag, not to be read as a precise dollar-to-condition-point conversion.
• 2025 is the newest FHWA data year available (published June 2025); IIJA's post-enactment window in this report is therefore under 4 years, shorter than the pre-law decade — the length-matched 2017-2021 comparison exists specifically to control for that asymmetry.

Every query behind this report

2 warehouse calls ran in this session, in order. Each is reproducible against the same snapshot.

SELECT "year", bridge_condition, COUNT(*) AS n
FROM transport.bridges
WHERE record_type = '1' AND "year" IN ('2021','2022','2023','2024','2025')
GROUP BY "year", bridge_condition
ORDER BY "year", bridge_condition

SELECT "year", bridge_condition, COUNT(*) AS n
FROM transport.bridges
WHERE record_type = '1' AND "year" IN ('2017','2018','2019','2020')
GROUP BY "year", bridge_condition
ORDER BY "year", bridge_condition