OBJECTIVE: We examined the prevalence of mental health problems in refugees living in camps that emerged in Tanzania during the Rwanda crisis that started in 1994. METHOD: Using the 28-item version of the General Health Questionnaire (GHQ-28), we examined two samples: a random sample (n = 854) and a sample of clients of a psychosocial support programme in these camps (n = 23). Sensitivity, specificity and positive- and negative predictive values were estimated for several cut-off scores of the GHQ-28. RESULTS: The prevalence of serious mental health problems was estimated at 50% (SE 12%). When using the GHQ-28 as a screener, a cut-off score of 14 is recommended. CONCLUSION: Given the high prevalence of mental health problems, psychosocial programmes for large refugee populations should aim at strengthening community structures and supporting groups instead of focusing at individuals. The screening capacity of the GHQ-28 could be used to identify mentally vulnerable groups.