OBJECTIVE Animal data suggest that cocaine has an immunosuppressive effect, but no human studies have been conducted to assess the relation of cocaine use with human papillomavirus (HPV) infection, the viral cause of cervical cancer. Since both cocaine use and HPV infection are common among HIV-positive women, we sought to determine whether use of cocaine and/or crack influences the natural history of HPV among women with or at high risk of HIV. METHODS Women enrolled in the Women's Interagency HIV Study (2278 HIV-seropositive and 826 high-risk seronegative women) were examined every six months for up to 9.5 years with Pap smear, collection of cervicovaginal lavage (CVL) samples, and detailed questionnaires regarding health and behavior, including use of crack and cocaine (crack/cocaine). CVLs were tested for HPV DNA by PCR, with genotyping for over forty HPV types. RESULTS In multivariate logistic regression models, censoring women treated for cervical neoplasia, crack/cocaine use within the last six months was associated with prevalent detection of oncogenic HPV DNA (odds ratio [OR] = 1.30 (1.09-1.55)), and with oncogenic HPV-positive squamous intraepithelial lesions (SIL) (OR = 1.70 (1.27-2.27)), following adjustment for age, race, HIV-serostatus, and CD4+ T-cell count, the number of sexual partners in the past six months, and smoking. In multivariate Cox models crack/cocaine use was also associated with a trend that approached significance in regard to incident detection of oncogenic HPV-positive SIL (HR = 1.51, 95% CI 0.99-2.30), and while the rate of oncogenic HPV clearance was not related to cocaine use, the clearance of any SIL was significantly lower in those with versus those without recent crack/cocaine use (HR = 0.57, 95% CI 0.34-0.97). CONCLUSIONS Cocaine use is associated with an increased risk of detection of both prevalent and incident oncogenic HPV infection, as well as an increased risk of HPV-positive SIL over time.