Background: Among women, trichomoniasis is the most common non-viral sexually transmitted infection worldwide, and is associated with serious reproductive morbidity, poor birth outcomes, and amplified HIV transmission. Single-dose metronidazole is the first-line treatment for trichomoniasis. However, bacterial vaginosis can alter treatment efficacy in HIV-infected women, and single-dose metronidazole treatment might not always clear infection. We compared single-dose metronidazole with a 7-day dose for the treatment of trichomoniasis among HIV-uninfected, non-pregnant women and tested whether efficacy was modified by bacterial vaginosis. Methods: In this multicentre, open-label, randomised controlled trial, participants were recruited at three sexual health clinics in the USA. We included women positive for Trichomonas vaginalis infection according to clinical screening. Participants were randomly assigned (1:1) to receive either a single dose of 2 g of metronidazole (single-dose group) or 500 mg of metronidazole twice daily for 7 days (7-day-dose group). The randomisation was done by blocks of four or six for each site. Patients and investigators were aware of treatment assignment. The primary outcome was T vaginalis infection by intention to treat, at test-of-cure 4 weeks after completion of treatment. The analysis of the primary outcome per nucleic acid amplification test or culture was also stratified by bacterial vaginosis status. This trial is registered with ClinicalTrials.gov, number NCT01018095, and with the US Food and Drug Administration, number IND118276, and is closed to accrual. Findings: Participants were recruited from Oct 6, 2014, to April 26, 2017. Of the 1028 patients assessed for eligibility, 623 women were randomly assigned to treatment groups (311 women in the single-dose group and 312 women in the 7-day-dose group; intention-to-treat population). Although planned enrolment had been 1664 women, the study was stopped early because of funding limitations. Patients in the 7-day-dose group were less likely to be T vaginalis positive at test-of-cure than those in the single-dose group (34 [11%] of 312 vs 58 [19%] of 311, relative risk 0·55, 95% CI 0·34–0·70; p<0·0001). Bacterial vaginosis status had no significant effect on relative risk (p=0·17). Self-reported adherence was 96% in the 7-day-dose group and 99% in the single-dose group. Side-effects were similar by group; the most common side-effect was nausea (124 [23%]), followed by headache (38 [7%]) and vomiting (19 [4%]). Interpretation: The 7-day-dose metronidazole should be the preferred treatment for trichomoniasis among women. Funding: National Institutes of Health.