OBJECTIVE: To describe the maternal acceptability of formula-feeding proposed to reduce postnatal HIV transmission in Abidjan, C?d'Ivoire. METHODS: Each consenting HIV-infected pregnant women, age > or =18 years, who received a perinatal antiretroviral prophylaxis was eligible. Two hierarchical infant-feeding options were proposed antenatally: exclusive formula-feeding or short-term exclusive breast-feeding. Formula-feeding was provided free up to age 9 months. Determinants of acceptability were analyzed using a logistic regression. Formula-feeding failure was defined as having breast-fed one's child at least once. RESULTS: Between March 2001 and March 2003, 580 women delivered: 97% expressed their infant-feeding choice before delivery; 53% chose formula-feeding. Significant prenatal determinants for refusing formula-feeding were: living with her partner, being Muslim, having a low educational level, being followed in one of the study sites, having not disclosed her HIV status, and having been included within the first 6 months of the project. Among the 295 mothers who formula-fed, the Kaplan-Meier probability of success of the formula-feeding option was 93.6% at Day 2 (95% confidence interval [CI]: 90.7% to 96.3%) and 84.2% at 12 months (95% CI: 79.9% to 88.5%): 46 of 295 (15.6%) women breast-fed at least once, of whom 41% temporarily practiced mixed-feeding at Day 2 because of social stigma or newborn poor health. CONCLUSIONS: In settings with general access to clean water, structured antenatal counseling, and sustained provision of free formula, slightly over half of HIV-infected women chose to artificially feed their newborn infant. Low mixed-feeding rates were observed. This social acceptability must be balanced with mother-child long-term health outcomes to guide safe recommendations on infant-feeding among HIV-infected women in African urban settings.