ENGLISH :
OBJECTIVE: This study examines the tendency toward problem behavior as an explanation for the relationship between problem substance use and occupational injury. METHOD: The authors used a matched case-control study nested in a cohort of 26,413 workers, in which cases (n = 3,994) were workers suffering an occupational injury. Five controls per case (n = 19,970) were selected from the cohort of workers active on the day of the injury and matched on job type. Conditional logistic regression modeled the association of problem substance use with occupational injury, controlling for problem behaviors and worker characteristics. Problem substance use was indicated indirectly if any of the following were alcohol/drug-involved during the comparison period: Employee Assistance Program visit, excused absence or disciplinary action. Discipline records identified minor (absenteeism) and serious (dishonesty, theft, assault, harassment, disrespect) problem behaviors during the comparison period. RESULTS: The odds of injury among workers with an indicator of problem substance use was 1.35 (p = .015) times greater than the odds among workers without an indicator, controlling for job type and demographics as well as adjusting for exposure. This ratio declined to 1.21 (p = .138) when problem behaviors were also controlled for. Minor and serious problem behaviors were significantly associated with occupational injury (odds ratio [OR] = 1.73, p