Most pages answering this question give you a number. This one gives you every number that is actually published, who published it, what sample it came from, and the specific thing it cannot tell you about your claim. Several of the most-quoted figures did not survive the check, and we say which ones and why.
Reviewed August 24, 2026 · verification policy · changelogNo government agency or insurance body publishes an average settlement. Anyone can check that. The word "settlement" appears once in the NAIC's 230-page auto insurance report, in a sentence with no figure attached to it, and never once on the Insurance Information Institute's auto statistics page. The Bureau of Justice Statistics put it in writing back in 1995: "details of tort settlements are unknown" and "little systematic data are available regarding why cases are settled or the cost of settlement for either party."
What does get published is the average insurance claim payment, plus verdict medians from tried cases. Those are real, sourced and useful if you know what they measure. They are also not settlements, and the gap between the two is where almost every misleading number on this topic comes from.
Seven figures clear our verification bar. Each one below was read at the primary source, and the last column is the part other sites leave off.
| Figure | What it measures | What it cannot tell you |
|---|---|---|
| $28,278 2024 data | Average paid auto bodily-injury liability claim Insurance Information Institute, from ISO (a Verisk business) | Paid claims only, so claims closed without payment are not in the denominator. Includes the insurer's loss adjustment expenses, so it is not what the claimant received. Excludes Massachusetts and most no-fault states. |
| $25,915.56 2022 data | Incurred cost per auto bodily-injury claim, countrywide excluding Texas and Massachusetts NAIC, 2022/2023 Auto Insurance Database Report (Table 8D) | An actuarial estimate of ultimate cost, including reserves for open and not-yet-reported claims, not money paid. Includes loss adjustment expenses. Insurers do not count claims the same way, which NAIC says makes state comparisons unreliable. |
| $12,174.94 2022 data | Incurred cost per personal injury protection (PIP) claim, excluding Texas NAIC, 2022/2023 Auto Insurance Database Report (Table 17D) | Only meaningful in the roughly two dozen states that write PIP, and Michigan alone runs more than five times the national figure, which drags the average up. |
| $2.00 per $1 mid-2017 to mid-2022 | Total settlement paid per dollar of medical bills on auto injury claims, up from $1.80 in early 2018 Insurance Research Council, Auto Injury Insurance Claims: A Study of Increasing Claim Severity | The closest thing to a published settlement ratio, but it comes from nine insurers covering about 43% of the market rather than a census, the window closed over four years ago, and IRC is funded by the insurers who supplied the data. |
| $24,000 2005 data | Median damage award to plaintiffs who won a tort jury trial (auto cases: $15,000) Bureau of Justice Statistics, Tort Bench and Jury Trials in State Courts, 2005 (NCJ 228129) | A verdict, not a settlement, in the 4% of tort cases that were tried, for the 51.6% of those plaintiffs who won. Measured before post-trial motions and appeals. Twenty-one years old. |
| $100,000 2014–2020 | Median personal-injury jury award reported to a commercial verdict database (mean: $1,479,368) Thomson Reuters, Current Award Trends in Personal Injury, 61st edition, reprinted by III | Verdicts voluntarily reported to a subscription database, so large plaintiff wins are over-represented. Excludes punitive damages. The full range runs from $1 to $740 million. III stopped reprinting the series after 2020. |
| $31,300 + $30,600 accident year 2025, preliminary | Average indemnity and medical cost per lost-time workers' compensation claim NCCI, State of the Line, Annual Insights Symposium 2026 | A projected ultimate cost per claim, not a settlement and not a payment. Lost-time claims only, so medical-only claims are excluded, in NCCI's ratemaking states only. |
NAIC prints two countrywide lines for bodily injury, and sites that quote it almost always take the bigger one. The bigger one is an artefact. Texas reports its incurred losses but does not report claim counts, so Texas money lands in the numerator while no Texas claims land in the denominator. Massachusetts bundles uninsured-motorist and medical-payments premiums into its bodily injury line, so NAIC strips it out too.
| Countrywide incurred BI losses, 2022 | Table 8B | $47,118,450,338 |
| Countrywide incurred BI claims | Table 8B | 1,629,333 |
| NAIC's headline 'Countrywide' severity | $47,118,450,338 ÷ 1,629,333 | $28,918.86 |
| But Texas reports those losses with no claim count | −$5,217,270,267 in losses, claims "n/a" | n/a |
| And Massachusetts bundles other coverages into BI | −$822,198,773 and −44,224 claims | excluded |
| NAIC's second line: countrywide without TX and MA | $41,078,981,298 ÷ 1,585,109 | $25,915.56 |
| The gap that Texas alone creates | $28,918.86 − $25,915.56 | $3,003.30 |
An 11.6% overstatement, entirely mechanical. NAIC publishes the second line precisely so readers can avoid the first, and states its Texas data limitation plainly in the technical notes. We use $25,915.56.
The Insurance Information Institute republishes its auto loss table every year on a rolling ten-year window, and the figures for a given year move as insurers report updated data. Bodily injury severity for 2020 was published as $20,235 in the 2011–2020 table and reads $20,769 in the current 2015–2024 table. The 2021 figure has moved from $22,734 to $23,592. Neither revision is improper, but it means any page quoting a III figure without naming the table vintage is quoting a number III no longer publishes. Ours comes from the 2015–2024 table, and we will say so each time it changes.
The attorney-versus-no-attorney comparison that circulates most widely, $77,600 with a lawyer against $17,600 without, is usually credited to an Insurance Research Council study called Paid in Full. We went looking for it. It is not in IRC's own publications index, across all nine pages of it, and we could not find it anywhere else either.
Those two numbers do have a real source. They are, to the dollar, the figures from the 2017 Martindale-Nolo reader survey published on Nolo's own site. A self-selected survey of one website's visitors has been re-badged as insurance-industry research under a title that does not appear to exist. That is worth knowing before you weigh any number in this topic, including ours.
BJS never published that. What BJS published is that bench and jury trials were about 4% of tort dispositions in 2005, and separately, for large-county tort cases in 1991–92, that 73% settled, 3% reached a trial verdict and 10% were dismissed. Those do not add to 100%, because roughly 14% resolved in other ways. The popular version quietly converts "did not reach a verdict" into "settled," which is the one thing the source is careful not to say.
We used to illustrate this with invented round numbers. We no longer need to, because the Thomson Reuters verdict series shows it directly:
| Median reported personal-injury jury award | 2014–2020 | $100,000 |
| Mean of the same awards | 2014–2020 | $1,479,368 |
| How far the mean sits above the median | $1,479,368 ÷ $100,000 | 14.8× |
| Middle half of all awards fell between | 25th–75th percentile | $20,000 – $505,000 |
| Largest single award in the series | top of range | $740,000,000 |
A mean nearly fifteen times the median describes the outliers, not the data. Every "average settlement is $X" headline you have read is a number shaped by cases that look nothing like a typical claim.
NAIC reports bodily injury cost per claim state by state for 2022, and the range runs from $13,684.96 in Arkansas to $77,661.00 in Michigan. North Carolina sits at $13,886.24, Ohio at $16,480.86, Florida at $28,242.31, California at $31,467.48, New York at $43,180.00. That is a 5.7-fold spread inside the same year and the same dataset, which is why we build state pages with each state's own rules rather than quote you a countrywide figure. NAIC adds a warning of its own: because some insurers count claims per accident and others per claimant, and nobody has measured the mix, "direct comparisons between state results could be misleading."
The BJS median gets quoted constantly as an average settlement, and it is close to the opposite. It counts only cases tried to completion, so every settled case is excluded by construction. About 4% of tort dispositions in the 2005 BJS sample reached trial, and 4% of civil dispositions in the National Center for State Courts' ten-county study. In the federal courts for the year ending September 2025 it was 1,681 of 407,379 civil terminations, or 0.41%, and that share keeps falling.
A verdict is also not money yet. BJS tracked appeals from civil trials and found plaintiff verdicts were reversed or modified 42% of the time, against 21% for defence verdicts. So even the tried-case median overstates what winning plaintiffs finally collected.
Our editorial policy is that a number we cannot trace does not get published, even when leaving it out costs us the easy answer. These did not make the page:
The arithmetic both sides of the table already run: your medical bills plus lost wages, times a severity multiplier, plus property damage, reduced by your share of fault. Every input is yours, every step is published on our methodology page, and the calculator prints each line so you can check it rather than believe it.
One published figure is worth carrying into that negotiation. IRC found auto bodily-injury claimants received about $2.00 in total settlement per dollar of medical bills, up from $1.80 in early 2018 to over $2.30 by mid-2022. Treat it as a sanity check on the order of magnitude, not a target: it comes from nine insurers rather than the whole market, the window closed in 2022, and IRC is funded by the carriers who supplied the data.
No government agency or insurance body publishes one, and that is checkable. The NAIC's 230-page auto report uses the word 'settlement' exactly once, in a sentence with no number attached, and the Insurance Information Institute's auto statistics page never uses it at all. What does get published is the average insurance claim payment, which is a different quantity, and the most-quoted 'average settlement' figures trace back to a reader survey whose publisher has never disclosed how many people answered it. The figures that do exist are all on this page with their limits stated.
The closest verifiable figure is the average paid auto bodily-injury liability claim, $28,278 for 2024 data, published by the Insurance Information Institute from ISO data. Read the footnotes before you use it: it counts only claims that were paid, it includes the insurer's own adjustment expenses rather than just the claimant's money, and it leaves out Massachusetts and most no-fault states. It is also a national figure covering states that range from about $13,700 to about $77,700 per claim.
A 2017 Martindale-Nolo survey of Nolo's own readers. Nolo still publishes it, and to Nolo's credit the page says openly that people with the most severe injuries 'immediately retained a lawyer and didn't participate in our survey.' What no version of it discloses is the number of respondents, which is why we treat it as an artefact rather than a statistic. Note also that $52,900 sits above the top of the $3,000 to $25,000 band where more than half of those readers landed, so the mean is being carried by a thin tail of large cases.
That claim circulates attributed to the Insurance Research Council, and IRC's own data points the other way. In its severity study, auto bodily-injury claimants with attorneys received $1.40 per dollar of medical expense after fees, against $1.80 for claimants without attorneys, and represented claimants waited a median of nearly 440 days to close. That is not an argument against hiring a lawyer, because IRC does not control for case severity and the harder, more serious, more contested claims are exactly the ones that get lawyers. It is an argument against the 3.5× number, which we could not source to any IRC publication.
No. NCCI, the workers' comp rating bureau, publishes average claim cost, about $31,300 in indemnity and $30,600 in medical per lost-time claim for accident year 2025. Those are projected ultimate costs for a claim, not negotiated settlement amounts, and they cover lost-time claims in NCCI's ratemaking states only. Any site presenting them as average settlements is relabelling them. Your own state's rates and schedules do the real work, and those are on each state's workers' comp page.
Mostly because they measure different things on different bases. For 2022 the Insurance Information Institute reports $24,681 per paid auto bodily-injury claim while the NAIC reports $25,915.56, and the two are not comparable: III divides paid losses by paid claims, NAIC divides incurred losses, including reserves for claims not yet reported, by incurred claims. Close enough to look interchangeable, different enough to be wrong if you treat them that way.
Your own numbers. Your documented medical bills and lost wages, a severity multiplier, your property damage, and your state's shared-fault rule. That is the arithmetic both adjusters and attorneys actually run, and our calculator prints every line of it so you can check the result rather than trust it.