OBJECTIVE: To assess the cost-effectiveness of alternative strategies of nevirapine (NVP) administration to prevent vertical HIV transmission in sub-Saharan Africa. DESIGN: A decision-analysis model was constructed to estimate the costs and effects of NVP-based prevention strategies for two separate groups of women: those who qualify for standard therapy by attending a 36-week prenatal visit, and those who do not qualify, owing to preterm delivery or lack of prenatal care. RESULTS: For women in prenatal care, mass provision of NVP without maternal serodiagnosis was found to yield greater health gains at an acceptable cost, compared with providing targeted therapy to only those women identified as seropositive. However, this conclusion was strongly contingent on several uncertain assumptions, most importantly the probability that a woman who does not know her serostatus will nonetheless adhere to therapy. Among those women who present for delivery without prior enrollment in a prenatal strategy, either late provision of maternal-infant NVP or treatment of only the infant would likely be a cost-effective alternative to the current practice of offering no preventive therapy. CONCLUSIONS: NVP intervention offers a cost-effective avenue for preventing vertical HIV transmission in sub- Saharan Africa. The optimal choice between mass therapy and targeted therapy cannot be confidently identified without information regarding adherence among women who do not know their serostatus. For women who do not receive NVP prenatally, treatment on presentation for delivery would be cost-effective even in the face of modest clinical efficacy. Clinical assessment of adherence to therapy among women who do not know their status and the field effectiveness of alternative approaches to NVP administration is urgently needed to allow identification of optimal prevention strategies.