From February 1992 until June 1994, all patients with histologically proven leprosy examined at the Leprology Unit of the Institut Marchoux in Bamako, Mali, were screened for HIV serology. In total, 740 leprosy patients have been tested; 553 known, previously treated leprosy cases and 187 new cases, mainly self-reporting and referred cases. The global seroprevalence in the sample was 1.5% (11/740), and increased from 1.3% in 1992 to 3.1% in 1994. HIV seroprevalence was higher in paucibacillary (PB) than in multibacillary (MB) cases (3.8% versus 0.8%, p < 0.05), and was slightly higher in new cases than in known, already treated cases (2.1% versus 1.3%), although not significantly. Among the 553 known, already treated leprosy patients, 1 out of 7 HIV-seropositive patients relapsed, as opposed to 34 out of 546 HIV-seronegative cases (14.2% versus 6.2%, p = 0.36). Among the new cases, none of the 37 patients with reaction and/or neuritis was HIV positive. In known, treated leprosy cases, there was no difference in the frequency of reactions and/or neuritis between HIV-positive and HIV-negative cases. Migration in a neighboring country appeared to be a risk factor for HIV seropositivity in our sample (chi 2 = 4.5, p = 0.04). In order to estimate the association of HIV with leprosy as compared to the general population, a control group of blood donors was set up, matched for age and sex. There was, however, no difference in HIV seroprevalence between the control group (9/735, 1.2%) and the leprosy group (1.5%). Although leprosy patients recruited for this study constitute a highly selected sample, it appears that HIV infection has little effect on leprosy, particularly on the PB/MB ratio, leprosy reactions and neuritis, but there is a suggestion the HIV infection might be associated with increased frequency of relapse.