BACKGROUND: There is insufficient evidence to support visceral leishmaniasis (VL) treatment recommendations in Brazil and an urgent need to improve current treatments. Drug combinations may be an option. METHODS: A multicenter, randomized, open label, controlled trial was conducted in five sites in Brazil to evaluate efficacy and safety of (i) amphotericin B deoxycholate (AmphoB) (1 mg/kg/day for 14 days), (ii) liposomal amphotericin B (LAMB) (3 mg/kg/day for 7 days) and (iii) a combination of LAMB (10 mg/kg single dose) plus meglumine antimoniate (MA) (20 mg Sb+5/kg/day for 10 days), compared to (iv) standard treatment with MA (20 mg Sb+5/kg/day for 20 days). Patients, aged 6 months to 50 years, with confirmed VL and without HIV infection were enrolled in the study. Primary efficacy endpoint was clinical cure at 6 months. A planned efficacy and safety interim analysis led to trial interruption. RESULTS: 378 patients were randomized to the four treatment arms: MA (n = 112), AmphoB (n = 45), LAMB (n = 109), or LAMB plus MA (n = 112). A high toxicity of AmphoB prompted an unplanned interim safety analysis and this treatment arm was dropped. Per intention-to-treat protocol final analyses of the remaining 332 patients show cure rates at 6 months of 77.5% for MA, 87.2% for LAMB, and 83.9% for LAMB plus MA, without statistically significant differences between the experimental arms and comparator (LAMB: 9.7%; CI95% -0.28 to 19.68, p = 0.06; LAMB plus MA: 6.4%; CI95% -3.93 to 16.73; p = 0.222). LAMB monotherapy was safer than MA regarding frequency of treatment-related adverse events (AE) (p = 0.045), proportion of patients presenting at least one severe AE (p = 0.029), and the proportion of AEs resulting in definitive treatment discontinuation (p = 0.003). CONCLUSIONS: Due to lower toxicity and acceptable efficacy, LAMB would be a more suitable first line treatment for VL than standard treatment.