A cross-sectional study of 854 outpatients seen at three hospitals in the Eastern Region of Ghana, West Africa assessed HIV-1 and HIV-2 seroprevalence rates and associated risk factors. During mid-1999, patients were tested in an anonymous study linked to a questionnaire and abstracted record data in three distinct communities with differing levels of urbanization. Combined HIV seroprevalence rates were 19.2% for Agomanya, the most rural community, 5.1% for Akwatia, and 3.4% for Nkawkaw, the most urbanized community. HIV-1 infection alone accounted for 84.5% of cases, compared with 9.5% and 6.0% for HIV-2 and dual infection, respectively. In a multivariable model, the Agomanya location was significantly associated with HIV status (odds ratio [OR], 7.7; 95% confidence interval [CI], 2.9-20.1; p <.0001). Employed people were also more likely to be infected (OR, 4.3; 95% CI, 1.02-18.3; p =.047). Among prenatal patients, those from Agomanya were more likely to report no live-in partner or spouse compared with Nkawkaw and Akwatia (p <.0001). These findings suggest sexual partner exchange patterns may be more useful in explaining HIV seroprevalence rates than level of community urbanization. Intervention programs directed at these high risk populations for HIV infection in Ghana are urgently needed.