We evaluated generic sodium stibogluconate (SSG) (International Dispensary Association, Amsterdam) versus Pentostam (sodium stibogluconate, GlaxoWellcome, London) under field conditions in Ethiopian patients with visceral leishmaniasis (VL; kala-azar). The 199 patients were randomly assigned to Pentostam (n = 104) or SSG (n = 95) in 1998/99; both drugs were given at 20 mg/kg intra-muscularly for 30 days. A clinical cure after 30-days treatment was achieved in 70.2% (Pentostam) and 81.1% (SSG). There were no significant differences between the 2 drugs for the following parameters: frequency of intercurrent events (vomiting, diarrhoea, bleeding or pneumonia) or main outcome (death during treatment and death after 6-month follow-up; relapse or post kala-azar dermal leishmaniasis at 6-months follow-up). Twenty-seven patients had confirmed co-infection with HIV. On admission, HIV co-infected VL patients were clinically indistinguishable from HIV-negative VL patients. The HIV co-infected VL patients had a higher mortality during treatment (33.3% vs 3.6%). At 6-month follow-up, HIV-positive patients had a higher relapse rate (16.7% vs 1.2%), a higher death rate during the follow-up period (14.3% vs 2.4%), and more frequent moderate or severe post kala-azar dermal leishmaniasis (27.3% vs 13.3%). Only 43.5% of the HIV-positive patients were considered cured at 6-months follow-up vs 92.1% of the HIV-negative patients. HIV-positive patients relapsing with VL could become a reservoir of antimonial-resistant Leishmania donovani.