Background: Nonoxynol-9 (rINN, nonoxinol-9) is an over-the-counter spermicide that has in-vitro anti-HIV-1 activity. Results of studies of its effectiveness in prevention of HIV-1 infection in women have been inconclusive. We aimed to assess effectiveness of this vaginal gel.Methods: We did a randomised, placebo-controlled, triple-blinded, phase 2/3 trial with COL-1492, a nonoxynol-9 vaginal gel, in 892 female sex workers in four countries: Benin, Cote d'lvoire, South Africa, and Thailand. 449 women were randomly allocated nonoxynol-9 and 443 placebo. Primary endpoint was incident HIV-1 infection. Secondary endpoints included Neisseria gonorrhoeae and Chlamydia trachomatis infections. Analysis was by intention to treat.Findings: 765 women were included in the primary analysis. HIV-1 frequency in nonoxynol-9 users was 59 (16%) of 376 compared with 104 (27%) of 389 in placebo users (402.5 vs 435.0 woman-years; hazard ratio adjusted for centre 1.5; 95% CI 1.0-2.2; p=0.047). 239 (32%) women reported use of a mean of more than 3.5 applicators per working day, and in these women, risk of HIV-1 infection in nonoxynol-9 users was almost twice that in placebo users (hazard ratio 1.8; 95% CI 1.0-3.2). 516 (68%) women used the gel less frequently tan 3.5 times a day, and in these, risk did not differ between he two treatments. No significant effect of nonoxynol-9 on N gonorrhoeae (1.2; 0.9-1.6) or C trachomatis (1.2; 0.8-1.6) infections was reported.Interpretation: This study did not show a protective effect of COL-1492 on HIV-1 transmission in high-risk women. Multiple use of nonoxynol-9 could cause toxic effects enhancing HIV-1 infection. This drug can no longer be deemed a potential HIV-1-prevention method. Assessment of other microbicides should continue.