We will conduct a community-based prospective cohort study that will offer home-based HIV counseling and testing to heterosexually active couples and obtain reproductive and sexual health data, and pregnancy and HIV incidence from men and women in Nyando district, western Kenya. A randomly-selected cohort of HIV discordant (n≅960), concordant-negative (subset of n≅960) and concordant-positive (n≅480) couples will be followed. Qualitative data will also be collected via dyad interviews (n=30) and 12 focus group discussions (among HIV-negative males, HIV-negative females, HIV-positive males, HIV-positive females), conducted at baseline and at month 18. We will apply the theoretical model elements to the cohort data using structural equation modeling to assess its value in predicting incident HIV and pregnancy. A policy meeting will be held in Kenya to discuss study findings. This study will contribute significantly to design of interventions to reduce horizontal and vertical HIV transmission in resource-limited settings.