A randomized un-blinded study on the treatment of oropharyngeal and esophageal candidiasis was conducted in Kinshasa (Zaire), among 141 inpatients with AIDS and oropharyngeal candidiasis, of whom 136 also had esophageal candidiasis. The study compared the efficacy of gentian violet mouth washes (1.5 ml 0.5% aqueous solution b.i.d.), oral ketoconazole (200 mg/day, after a meal) and nystatin mouth washes (200.000 U oral suspension q.i.d.). Patients treated with mouth washes swallowed their medication after mouth washing. Patients enrolled in this study had a very high mortality (probability of death: 41.6% after 14 days). After 14 days, 72 patients could be evaluated. At that time, oropharyngeal lesions had disappeared in similar proportions of patients treated with gentian violet (11/26, 42%) and ketoconazole (10/23, 43%), and in a lower proportion of patients treated with nystatin (2/23, 9%; p less than 0.05). In esophageal candidiasis, ketoconazole seemed more efficient than both other treatments: esophageal lesions had disappeared in 5 (24%) of the 21 patients on ketoconazole, compared to less than 10% of patients on both other treatments (p = 0.07). The suboptimal results observed with all 3 treatments could be explained by the profound immunosuppression of patients enrolled in the study. This study suggests that gentian violet is effective treatment for oropharyngeal candidiasis. As it is very cheap (0.5 US$/treatment course in Kinshasa), we suggest that its use should be assessed in larger studies.