Objective: Injured drivers with blood alcohol concentration (BAC) above the legal limit are rarely convicted of impaired driving. One explanation is that police may have difficulty recognizing alcohol intoxication in injured drivers. In this study we compare police documentation of alcohol involvement with BAC measured on arrival in hospital. Our objectives are to determine how often police document alcohol involvement in injured drivers with BAC > 0.05%, and identify factors that influence police documentation of alcohol involvement. Methods: We included injured drivers (1999 – 2003) who were admitted to a British Columbia (BC) trauma centre or treated in the Vancouver General Hospital emergency department. We used probabilistic linkage to obtain police collision reports. Police were considered to indicate alcohol involvement if i) police documented that alcohol contributed to the crash, ii) the driver received an administrative sanction for impaired driving, or iii) the driver was criminally convicted of impaired driving. The proportions of drivers for whom police indicated alcohol involvement was determined relative to age, gender, BAC levels, crash severity and crash characteristics. Multivariate logistic regression was used to identify factors independently associated with police indication of alcohol involvement. Results: 2410 injured drivers (73.5% male) were matched to a police report. Overall, 857 (35.6%) drivers tested positive for alcohol (BAC > 0) and 736/857 (85.9%) of alcohol positive drivers had a BAC > 0.05% (the BC legal limit). Of the 736 drivers with BAC > 0.05% at time of admission, police indicated alcohol involvement in 530 (72.0%). The criminal code conviction rate for impaired driving was 4.7% for drivers with 0.08% ≤ BAC < 0.16%, and 13.6% for drivers with BAC > 0.16%. The following factors were associated with higher odds of police indicating alcohol involvement: i) increasing blood alcohol levels, ii) a prior record of impaired driving, iii) involvement in a single vehicle crash, iv) involvement in a night time crash, and v) traffic violations or unsafe driving actions recorded by police. Conclusions: Police recognized and documented alcohol involvement in 72% of injured drivers with BAC ≥ 0.05%. Police documentation of alcohol involvement is more common at higher BAC levels, in night time or single vehicle crashes, for drivers who committed traffic violations or drove unsafely, and for drivers with a prior record of impaired driving. The low conviction rate of injured impaired drivers does not appear to be due to police inability to recognize alcohol involvement.