OBJECTIVE: To measure trends in adult mortality in countries with significant levels of HIV prevalence using data sources other than those that collect information on HIV status. DATA AND METHODS: Data sources consisted of national population censuses and sample surveys, Demographic and Health Surveys (DHS), vital registration, and longitudinal surveillance systems. Estimates of adult mortality were derived from censuses using intercensal survival and questions on deaths of household members and orphanhood. From DHS adult mortality was measured from data on survival of siblings and orphanhood. Death registration should be tested for changes in the level of coverage before drawing conclusions on mortality trends. Demographic surveillance systems record trends and age patterns of mortality, but are not nationally representative. RESULTS: Census and survey data from Kenya, Malawi and Zimbabwe showed increasing adult mortality in the 1990s, reversing previous downward trends. DHS data for over 20 sub-Saharan countries showed that most had increasing mortality, which was steepest in eastern and southern Africa, with high HIV prevalences. Death registration in Zimbabwe and South Africa showed increasing adult mortality, as in Thailand and Trinidad, both countries with appreciable levels of HIV. Surveillance systems in Tanzania and South Africa showed radically different age patterns of mortality, with relatively high rates among younger adults compared with data from countries with lower HIV prevalences. CONCLUSION: Adult mortality is increasing in countries with high HIV/AIDS prevalences, although the contribution of the epidemic to this increase is difficult to measure. More data and improved methods of analysis are needed before firm conclusions can be drawn.