AIMS:To conduct a cost-effectiveness analysis (CEA) comparing the delivery of brief intervention (BI) with brief treatment (BT) within Screening, Brief Intervention and Referral to Treatment (SBIRT) programs. DESIGN:Quasi-experimental differences in observed baseline characteristics between BI and BT patients were adjusted using propensity score techniques. Incremental comparison of costs and health outcomes associated with BI and BT. SETTING:Health-care settings in four US states participating in Substance Abuse and Mental Health Services Administration SBIRT grant programs. PARTICIPANTS:Ninety patients who received BT and 878 who received BI. MEASUREMENTS:Per-patient cost of SBIRT, patient demographics and six measures of substance use: proportion using alcohol, proportion using alcohol to intoxication, days of alcohol use, days of alcohol use to intoxication, proportion using drugs and days using drugs. FINDINGS:BI and BT were associated with better outcomes. The cost of SBIRT was significantly higher for BT patients ($75.54 versus 16.32, 95% confidence interval, P