In HIV cohorts in sub-Saharan Africa, documenting vital status of patients lost to follow-up is a major challenge. The effect of specific vital status investigation procedures (VSIPs) on the number of known deaths has never been shown. We assessed the effects of VSIP on survival estimates in a 4-year prospective cohort study in Abidjan, Côte d'Ivoire. As of June 2000, 545 HIV-infected adults had been followed for 1186 person-years, of whom 233 were documented as deceased. Forty-eight percent of deaths were known through scheduled VSIPs, including reading of the newspaper obituaries (2%), telephone calls to relatives (10%), and home visits (36%). Survival probability at 1, 2, and 3 years was estimated to be 0.79, 0.65, and 0.56, respectively. Without VSIP, survival at 1, 2, and 3 years would have been estimated to be 11, 23, and 30% higher, respectively. In this large African capital city, survival estimates closely depended on VSIPs, mainly home visits. We suggest that the percentage of deaths known through VSIPs would be a useful indicator to be added when reporting survival data from urban HIV cohort studies in sub-Saharan Africa.