DESIGN: We estimated the cost-effectiveness ratios of three treatments for heroin dependence. We used a microcosting methodology to determine fixed, variable, and societal costs of each intervention. Cost data were collected from investigators, staff, and project records on the number and type of resources used and unit costs; societal costs for participants' time were estimated using Malaysia's minimum wage. Costs were estimated from a provider and societal perspective and reported in 2004 US dollars.

SETTING: Muar, Malaysia.

PARTICIPANTS: 126 patients enrolled in a randomized, double-blind, placebo-controlled clinical trial in Malaysia (2003-2005) receiving counseling and buprenorphine, naltrexone, or placebo for treatment of heroin dependence.

MEASUREMENTS: Primary outcome measures included days in treatment, maximum consecutive days of heroin abstinence, days to first heroin use, and days to heroin relapse. Secondary outcome measures included treatment retention, injection drug use, illicit opiate use, AIDS Risk Inventory total score, and drug risk and sex risk subscores.

FINDINGS: Buprenorphine was more effective and more costly than naltrexone for all primary and most secondary outcomes. Incremental cost-effectiveness ratios were below $50 for primary outcomes, mostly below $350 for secondary outcomes. Naltrexone was dominated by placebo for all secondary outcomes at almost all endpoints. Incremental treatment costs were driven mainly by medication costs, especially the price of buprenorphine.