Background Human African trypanosomiasis (HAT; sleeping sickness) caused by Trypanosoma brucei gambiense is a fatal disease. Current treatment options for patients with second-stage disease are toxic, ineffective, or impractical. We assessed the efficacy and safety of nifurtimox-eflornithine combination therapy (NECT) for second-stage disease compared with the standard eflornithine regimen. Methods A multicentre, randomised, open-label, active control, phase III, non-inferiority trial was done at four HAT treatment centres in the Republic of the Congo and the Democratic Republic of the Congo. Patients aged 15 years or older with confirmed second-stage T b gambiense infection were randomly assigned by computer-generated randomisation sequence to receive intravenous eflornithine (400 mg/kg per day, every 6 h; n=144) for 14 days or intravenous eflornithine (400 mg/kg per day, every 12 h) for 7 days with oral nifurtimox (15 mg/kg per day, every 8 h) for 10 days (NECT; n=143). The primary endpoint was cure (defined as absence of trypanosomes in body fluids and a leucocyte count ≤20 cells per μL) 18 months after treatment. Efficacy analyses were done in the intention-to-treat (ITT), modified ITT, and per-protocol (PP) populations. The non-inferiority margin for the difference in cure rates was defined as 10%. This study is registered with ClinicalTrials.gov, number NCT00146627. Findings One patient from the eflornithine group absconded after receiving the first dose, without any type of assessment done, and was excluded from all analyses. In the ITT population, 131 (91·6%) of 143 patients assigned to eflornithine and 138 (96·5%) of 143 patients assigned to NECT were cured at 18 months (difference −4·9%, one-sided 95% CI −0·3; p<0·0001). In the PP population, 122 (91·7%) of 133 patients in the eflornithine group and 129 (97·7%) of 132 in the NECT group were cured at 18 months (difference −6·0%, one-sided 95% CI −1·5; p<0·0001). Drug-related adverse events were frequent in both groups; 41 (28·7%) patients in the eflornithine group and 20 (14·0%) in the NECT group had major (grade 3 or 4) reactions, which resulted in temporary treatment interruption in nine and one patients, respectively. The most common major adverse events were fever (n=18), seizures (n=6), and infections (n=5) in the eflornithine group, and fever (n=7), seizures (n=6), and confusion (n=2) in the NECT group. There were four deaths, which were regarded as related to study drug (eflornithine, n=3; NECT, n=1). Interpretation The efficacy of NECT is non-inferior to that of eflornithine monotherapy. Since this combination treatment also presents safety advantages, is easier to administer (ie, infusion every 12 h for 7 days vs every 6 h for 14 days), and potentially protective against the emergence of resistant parasites, it is suitable for first-line use in HAT control programmes. Funding Médecins Sans Frontières (Dutch section), Médecins Sans Frontières International, and the Drugs for Neglected Diseases Initiative.