RoadBacked Research · 2026
Where Do Passenger Vehicles Get Hit?
A National Analysis of Initial Impact Locations Among Passenger Vehicles in Police-Reported U.S. Crashes, 2020–2024
Published · By the RoadBacked Editorial Team · RoadBacked's analysis of NHTSA Crash Report Sampling System (CRSS) data.
Methodology at a glance
- Source
- NHTSA CRSS public microdata, 2020–2024
- Unit of analysis
- Passenger-vehicle involvement in a police-reported crash
- Measure
- IMPACT1_IM, area of impact / initial contact point
- Estimation
- CRSS WEIGHT, Taylor-linearized ratio estimates, 95% CIs
- Analytic records
- 409,115 sampled vehicle records
- Scope
- Descriptive estimates. No causal claims.
Executive findings
18.8%
Nearly 1 in 5 directional initial contacts occurred on the left or right side of a passenger vehicle (95% CI 18.3%–19.4%).
27.6%
Share of directional initial contacts that were side contacts at crashes coded as occurring at an intersection (95% CI 26.9%–28.4%), compared with 17.7% at non-junction locations (95% CI 17.1%–18.2%). Descriptive only.
18.4–19.5%
The estimated annual side-contact share stayed within this range in every year analyzed, 2020 through 2024.
Important limitation
Initial impact location does not measure dashcam visibility. A passenger vehicle can be struck on its side while a forward-facing camera still records important pre-impact movement. These results describe where vehicles were first contacted, not which camera configuration would or would not have captured useful evidence.
Why this question matters
Crash reporting usually describes what happened — a rear-end collision, an angle crash at an intersection, a sideswipe. Less often published is a simple national picture of where on the vehicle the first contact occurred, limited to passenger vehicles and expressed as a share of directional contacts.
That picture is useful to researchers, journalists, insurers, fleet and rideshare operators, and to drivers deciding how to document their vehicle. RoadBacked's analysis of NHTSA CRSS data estimates the distribution below. It is descriptive: it reports patterns in recorded police crash reports, not the causes behind them.
Where passenger vehicles are first contacted
Among directional initial contacts in the analyzed passenger-vehicle records, front contact was most common at 55.9%, followed by rear contact at 25.2%. Left-side and right-side contacts were close to each other, and together account for 18.8% of directional initial contacts (95% CI 18.3%–19.4%).
Where passenger vehicles are first contacted, 2020–2024
Share of directional initial contacts among passenger-vehicle involvements. Combined left and right side: 18.8%. Source: RoadBacked analysis of NHTSA CRSS 2020–2024.
| Initial contact | Estimate | 95% CI |
|---|---|---|
| Front | 55.9% | 55.5%–56.4% |
| Rear | 25.2% | 24.8%–25.7% |
| Left side | 9.5% | 9.2%–9.8% |
| Right side | 9.3% | 9.0%–9.6% |
| Either side | 18.8% | 18.3%–19.4% |
Side contact by road-junction context
At crashes coded as occurring at an intersection, 27.6% of directional initial contacts were side contacts, compared with 17.7% at non-junction locations. This is a descriptive difference between two reported road contexts. It does not establish that intersections cause side contacts, and the two groups differ in many unmeasured ways.
Side initial contact by road-junction context, 2020–2024
Share of directional initial contacts that were left- or right-side contacts, with 95% confidence intervals. Source: RoadBacked analysis of NHTSA CRSS 2020–2024.
| Road context | Estimate | 95% CI |
|---|---|---|
| At an intersection | 27.6% | 26.9%–28.4% |
| Non-junction | 17.7% | 17.1%–18.2% |
Side contact by vehicle type
Side initial-contact shares were similar across passenger-vehicle types. These are descriptive estimates with overlapping confidence intervals; differences between vehicle types are not automatically statistically significant, and nothing here implies that vehicle design produces the pattern.
Side initial contact by vehicle type, 2020–2024
Share of directional initial contacts that were left- or right-side contacts, with 95% confidence intervals. Source: RoadBacked analysis of NHTSA CRSS 2020–2024.
| Vehicle type | Estimate | 95% CI |
|---|---|---|
| Passenger cars | 18.8% | 18.2%–19.4% |
| SUVs / utility | 18.0% | 17.4%–18.6% |
| Pickups | 19.7% | 18.9%–20.5% |
| Vans / minivans | 19.2% | 18.4%–19.9% |
Supplemental: urban and rural
Estimates by land use were close together and are reported here only for completeness.
| Land use | Estimate | 95% CI |
|---|---|---|
| Urban | 18.7% | 18.1%–19.2% |
| Rural | 19.3% | 17.7%–21.0% |
Five-year consistency, 2020–2024
The estimated annual side-contact share stayed within a narrow band in every year analyzed, from 19.5% in 2020 to 18.4% in 2024.
Annual side-contact share, 2020–2024
Share of directional initial contacts that were left- or right-side contacts, by year, with 95% confidence intervals. The estimates overlap; RoadBacked does not interpret this series as evidence of a clear downward trend. Source: RoadBacked analysis of NHTSA CRSS 2020–2024.
| Year | Estimate | 95% CI |
|---|---|---|
| 2020 | 19.5% | 18.2%–20.8% |
| 2021 | 18.8% | 17.5%–20.2% |
| 2022 | 19.2% | 17.6%–20.7% |
| 2023 | 18.4% | 17.3%–19.5% |
| 2024 | 18.4% | 17.3%–19.5% |
What this research does not show
- It does not show which crashes a camera would or would not have recorded. Initial contact location is not a measure of camera coverage or footage usefulness.
- It does not establish fault, causation or liability in any crash.
- It does not show that intersections cause side contacts, or that any vehicle type is safer or more exposed than another.
- It does not show a rising or falling trend in side contacts. The annual estimates overlap within their confidence intervals.
- It does not count unique crashes or unique vehicles. Weighted totals are sums of annual vehicle-involvement estimates.
- It is not an NHTSA publication or an NHTSA finding. It is RoadBacked's analysis of public NHTSA CRSS microdata.
Methodology
Source. NHTSA Crash Report Sampling System (CRSS), annual public CSV microdata for 2020, 2021, 2022, 2023 and 2024. CRSS is a nationally representative probability sample of police-reported crashes.
Unit of analysis. A passenger-vehicle involvement in a police-reported crash, not a unique crash. A crash involving two passenger vehicles contributes two records.
Passenger vehicles. Defined using NHTSA's BODY_TYP passenger-vehicle classification — passenger cars plus light trucks and vans — with unknown truck type included only under NHTSA's TOW_VEH condition. This is NHTSA's classification, not a custom RoadBacked grouping.
Initial contact variable. IMPACT1_IM, NHTSA's imputed "Area of Impact, Initial Contact Point."
Direction mapping. Front = 11, 12 and 1 o'clock. Rear = 5, 6 and 7 o'clock. Left side = 8, 9 and 10 o'clock plus left-side codes 61–63. Right side = 2, 3 and 4 o'clock plus right-side codes 81–83.
Exclusions. Non-directional categories such as non-collision, top, undercarriage and set-in-motion events are excluded from the front/rear/left/right percentage denominator. They represent about 1.25% of weighted passenger-vehicle involvements in this analysis. Records meeting the broad passenger-vehicle definition before directional-contact exclusions: 414,267 sampled vehicle records; directional-contact analytic records: 409,115.
Estimation. Estimates use the CRSS WEIGHT variable. 95% confidence intervals use the CRSS complex survey structure with PSU_VAR and PSUSTRAT, with survey strata and PSUs treated as year-specific when pooling 2020–2024, and Taylor-linearized ratio estimates.
Pooling. The pooled result combines annual weighted vehicle-involvement estimates. Summed annual weighted passenger-vehicle involvements: 47,831,619; directional-contact involvements: 47,233,260; side-contact involvements: 8,892,877. These are vehicle involvements across annual estimates, not unique crashes and not unique vehicles over five years.
Error. CRSS is a sample, so every estimate here is subject to sampling error, reflected in the confidence intervals, and to nonsampling error such as police reporting and coding practice, missing data and imputation.
Period. The report focuses on 2020–2024, the latest five full CRSS years available in this analysis. Earlier CRSS years remain valid data; they are simply outside this report's scope.
Data definitions and impact-code map
| Direction | IMPACT1_IM codes |
|---|---|
| Front | 11, 12, 1 o'clock |
| Rear | 5, 6, 7 o'clock |
| Left side | 8, 9, 10 o'clock, plus left-side codes 61–63 |
| Right side | 2, 3, 4 o'clock, plus right-side codes 81–83 |
| Excluded (non-directional) | Non-collision, top, undercarriage and set-in-motion events |
Download the aggregated results
The CSV contains only RoadBacked's aggregated estimates and methodology fields. It does not redistribute NHTSA microdata; the raw CRSS files are available directly from NHTSA.
Download aggregated results (CSV)Estimates, sample records, weighted denominators and numerators, and 95% confidence intervals for every segment in this report, at full precision.- Percentages are survey-weighted.
- sample_records are unweighted analytic records for the listed domain.
- Weighted counts are sums of annual CRSS vehicle-involvement estimates, not unique crashes.
- Confidence intervals are 95%.
- Values may not sum exactly due to rounding.
Sources
How to cite this research
RoadBacked. (2026). Where Do Passenger Vehicles Get Hit? A National Analysis of Initial Impact Locations Among Passenger Vehicles in Police-Reported U.S. Crashes, 2020–2024. RoadBacked Research.