OBJECTIVE The objective of this study was to compare human papillomavirus (HPV) DNA testing between self-administered vaginal swabs and physician-administered cervical swabs in women from rural Rakai District, Uganda. METHODS Between 2002 and 2003, women from a population-based cohort participated in an HPV study. Women collected self-administered vaginal swabs and were also offered a pelvic examination, which included physician-collected cervical samples. METHODS Hybrid-capture 2 was used to determine carcinogenic HPV status. Polymerase chain reaction was used to determine HPV genotypes. Unweighted kappa statistics were used to determine agreement. RESULTS Compliance with self-collected swabs was > or =86%; however, only 51% accepted a pelvic examination. Carcinogenic HPV prevalence was 19% in self-collected and 19% in physician-collected samples. Agreement among paired observations was 92% with a kappa of 0.75. Kappa between self- and physician-collected samples was similar in HIV strata (k = 0.71 and 0.75 for HIV-positive and HIV-negative, respectively). CONCLUSIONS In this community-based setting, detection of carcinogenic HPV was comparable among self- and physician-administered samples. Self-collection is a feasible and accurate means of obtaining HPV samples from women in resource-poor settings or persons reluctant to undergo a pelvic examination.