OBJECTIVE: To examine historical estimates of infant and under-five mortality in Afghanistan, provide estimates for rural areas from current population-based data, and discuss the methodological challenges that undermine data quality and hinder retrospective estimations of mortality. METHODS: Indirect methods of estimation were used to calculate infant and under-five mortality from a household survey conducted in 2006. Sex-specific differences in underreporting of births and deaths were examined and sensitivity analyses were conducted to assess the effect of underreporting on infant and under-five mortality. FINDINGS: For 2004, rural unadjusted infant and under-five mortality rates were estimated to be 129 and 191 deaths per 1000 live births, respectively, with some evidence indicating underreporting of female deaths. If adjustment for underreporting is made (i.e. by assuming 50% of the unreported girls are dead), mortality estimates go up to 140 and 209, respectively. CONCLUSION: Commonly used estimates of infant and under-five mortality in Afghanistan are outdated; they do not reflect changes that have occurred in the past 15 years or recent intensive investments in health services development, such as the implementation of the Basic Package of Health Services. The sociocultural aspects of mortality and their effect on the reporting of births and deaths in Afghanistan need to be investigated further.