Buprenorphine and methadone are the two established substitution drugs licensed in many countries for the treatment of opioid dependence. Little is known, however, about how these two drugs are applied and how they work in clinical practice. In this paper we present the aims, methods, design and sampling issues of a collaborative multi-stage epidemiological study (COBRA) to address these issues. Based on a nationally representative sample of substitution physicians, the study is designed as an observational, naturalistic study, consisting of three major parts. The first part was a national survey of substitution doctors (prestudy, n = 379 doctors). The second part was a cross-sectional study (n = 223 doctors), which consisted of a target-week assessment of 2,694 consecutive patients to determine (a) the severity and problem profiles and treatment targets; (b) the choice and dosage scheme of the substitution drug; (c) past and current interventions, including treatment of comorbid hepatitis C; and (d) cross-sectional differences between the two drugs with regard to comorbidity, clinical course, acceptance/compliance and social integration. The third part consists of a prospective-longitudinal cohort study of 48 methadone-treated and 48 buprenorphine-treated patients. The cohort is followed up over a period of 12 months to investigate whether course and outcome of the patients differ by type or treatment received in terms of clinical, psychosocial, pharmaco-economic and other related measures. The response rate among substitution doctors was 57.1%; that among eligible patients was 71.7%. Comparisons with the federal registers reveal that the final samples of doctors and patients may be considered nationally representative with regard to regional distribution, training, type of setting as well as the frequency of patients treated with buprenorphine or methadone. The COBRA study provides a unique comprehensive database, informing about the natural allocation and intervention processes in routine care and about the course and outcome of patients treated with buprenorphine or methadone.