Increasing knowledge about effective therapies for alcohol dependence calls for new research designs to examine treatment interactions between pharmacotherapies and behavioral interventions. In 1997, the National Institute on Alcohol Abuse and Alcoholism recruited 11 sites and a coordinating center for a large-scale (1,375 subjects), randomized placebo controlled trial to test 16 weeks of active treatment using naltrexone and acamprosate alone and in combination. Most participants receive 9 brief sessions delivered by medically trained providers to promote sobriety and enhance medication adherence (Medical Management, MM). Half the participants are also randomized to individualized psychotherapy (up to 20 sessions of Combined Behavioral Intervention, CBI), integrating elements of the successful behavioral interventions from Project MATCH. COMBINE seeks to evaluate the efficacy of the two most promising medications (naltrexone and acamprosate) both singly and together, when combined with different intensities of behavioral therapies. COMBINE incorporates a number of innovative design aspects, including a no-pill psychotherapy-alone condition, behavioral interventions that are both manual-guided and individualized, and pharmacotherapy dosing that is greater than in some previous trials. Two COMBINE pilot studies demonstrate the safety and acceptability of the combination pharmacotherapy dosing, and the feasibility of implementing the manualized behavioral interventions. This paper introduces COMBINE's goals, methods and analytic strategies, and their potential to improve multimodal treatment selection.