The present study compared in a clinical non-experimental setting the efficacy of buprenorphine (BUP) and methadone (METH) in the treatment of opioid dependence: all the subjects included in the study showed severe long-lasting heroin addiction. Participants (154) were applicants to a 12 weeks treatment program, who were assigned to either METH (78) (mean doses 81.5 +/- 36.4 mg) or BUP (76) (mean doses 9.2 +/- 3.4 mg) treatment. Aim of the study was to evaluate patient/treatment variables possibly influencing retention rate, abstinence from illicit drugs and mood changes. METH patients showed a higher retention rate at week 4 (78.2 versus 65.8) (P < 0.05), but BUP and METH were equally effective in sustaining retention in treatment and compliance with medication at week 12 (61.5 versus 59.2). Retention rate was influenced by dose, psychosocial functioning and not by psychiatric comorbidity in METH patients. In contrast, BUP maintained patients who completed the observational period showed a significantly higher rate of depression than those who dropped out (P < 0.01) and the intention to treat sample (P < 0.05). No relationship between retention and dose, or retention and psychosocial functioning was evidenced for BUP patients. The risk of positive urine testing was similar between METH and BUP, as expression of illicit drug use in general. At week 12, the patients treated with METH showed more risk of illicit opioid use than those treated with BUP (32.1% versus 25.6%) (P < 0.05). Negative urines were associated with higher doses in both METH and BUP patients. As evidenced for retention, substance abuse history and psychosocial functioning appear unable to influence urinalyses results in BUP patients. Buprenorphine maintained patients who showed negative urines presented a significantly higher rate of depression than those with positive urines (P < 0.05). Alternatively, psychiatric comorbidity was found unrelated to urinalyses results in METH patients. Our data need to be interpreted with caution because of the observational clinical methodology and non-random procedure. The present findings provide further support for the utility of BUP in the treatment of opioid dependency and demonstrate efficacy equivalent to that of METH during a clinical procedure. BUP seems to be more effective than METH in patients affected by depressive traits and dysphoria, probably due to antagonist action on kappa-opioid receptors. Psychosocial functioning and addiction severity cannot be used as valuable predictors of BUP treatment outcome. High doses appear to predict a better outcome, in term of negative urines, for both METH and BUP, but not in term of retention for BUP patients.