OBJECTIVE The authors analysed data from female sex workers screened prior to participation in a microbicide trial to examine the association between prevalent vaginal flora abnormalities and HIV infection, with special emphasis on the role of the intermediate vaginal flora (IVF) in this association. METHODS Data from the Kampala, Cotonou, Chennai and Mudhol/Jamkhandi sites were analysed. Participants were interviewed and provided blood for HIV and syphilis antibody testing, genital samples for the diagnosis of vaginal flora abnormalities (using Nugent score) and other reproductive tract infections. Log-binomial regression was used to estimate the HIV prevalence ratio (PR) in relation to IVF and bacterial vaginosis (BV). RESULTS Among 1367 women, BV, IVF and HIV prevalences were 47.6% (95% CI=45.0% to 50.3%), 19.2% (95% CI=17.1% to 21.2%) and 27.0% (95% CI=24.6% to 29.3%), respectively. In multivariate analysis, adjusting for study site, age, years of education, occupation, female sterilisation, oral sex, past history of sexually transmitted infection, gonorrhoea and candidiasis, IVF was significantly associated with HIV infection with a PR similar to that of BV (adjusted PR=1.56 (95% CI=1.22 to 1.98) and 1.48 (95% CI=1.20 to 1.84), respectively). CONCLUSIONS Though the cross-sectional design of the study precludes directional interpretation of the findings, the data do suggest that IVF may be as important as BV in HIV acquisition. The authors recommend prospective research to better understand the association between IVF and HIV acquisition.