This study aims to evaluate the discordance between police reports of injury severity among road casualties and the length of hospital stay and the Injury Severity Scale (ISS) by linking information from the crash records of the Hong Kong Police with the trauma records of a regional hospital. Sensitivity and specificity analyses suggest that police injury grading diverges noticeably from the definition of a 12-h hospital stay. Police reports overestimate injury severity remarkably. The results of logistic regression indicate that age, the ISS, and the position of the victim significantly determine the likelihood of police injury misclassification. Furthermore, an optimal demarcation point of the length of hospital stay for serious injury is estimated.