This paper examines the effects of antiretroviral therapy (ART) on demand for HIV testing and of ART-induced testing on demand for risky sexual behavior. I provide a model of sexual behavior decision-making under uncertainty and estimate the structural parameters of the model using nationally representative survey data from Zambia on HIV testing decisions before and after the introduction of ART. The empirical results indicate that although the introduction of ART appears to have increased demand for HIV testing, the ART allocation process may limit the prevention benefit of ART-induced testing. Simulation results show that eliminating this prevention inefficiency while holding the supply of ART constant would increase the prevention impact of ART-induced testing more than four-fold. More generally, the analysis indicates that existing studies which examine \universal" testing or quasi-experimental testing programs understate the efficacy of standard voluntary counseling and testing programs.