HIV prevalence is the most commonly used measure to prioritize communities for HIV prevention. We show that data on two HIV infection stages (early vs. nonearly and late vs. nonlate) allow estimation of two better measures of prevention need: HIV incidence (for prevention of HIV acquisition) and expected probability of HIV transmission in unprotected sex acts between HIV-infected community members and susceptible individuals (for prevention of HIV transmission). The three ranking schemes--by prevalence, incidence, and transmission probability--lead to significantly different community rank orders. Disease stage information should be collected in HIV surveys.