BACKGROUND: Several randomised trials have compared interferon-alpha with control as adjuvant therapy for high-risk malignant melanoma. The results of the individual trials have been either inconclusive or even apparently conflicting. To assess all the available evidence we performed a meta-analysis of these trials. METHODS: Standard methods for quantitative meta-analysis based on published data were used. Endpoints evaluated were recurrence-free survival and overall survival. A subgroup analysis by dose of interferon-alpha was performed. FINDINGS: Twelve trials, comprising 14 comparisons of interferon-alpha with control, with results available were identified. Recurrence-free survival was improved with interferon-alpha: hazard ratio 0.83, 95% confidence interval 0.77 to 0.90, p=0.000003. The benefit on overall survival was less clear (0.93, 0.85 to 1.02, p=0.1) and the confidence interval is compatible both with no benefit and with a moderate, but clinically worthwhile, benefit. There was some evidence of a dose response relationship with a significant trend for the benefit of interferon-alpha to increase with increasing dose for recurrence-free survival (test for trend: p=0.02) but not for overall survival (trend: p=0.8). INTERPRETATION: This meta-analysis provides the most reliable synthesis of the data currently available. Adjuvant interferon-alpha produces clear reductions in recurrence of high-risk melanoma, with some evidence of an effect of dose of interferon-alpha, but it is unclear whether this translates into a worthwhile survival benefit or not. Additional and more mature data are needed to resolve these issues and an individual patient data meta-analysis should be performed.