The premature death of adults is a major, but poorly documented, health problem in developing countries. The inadequacy of registration statistics and difficulty of collecting accurate data directly in surveys mean that indirect methods of analysis, particularly those based on orphanhood, represent an important source of adult mortality estimates. Assessments of the orphanhood method have expressed concern about the robustness of the procedures used to estimate life table indices from orphanhood data, particularly for males, about under-reporting of orphanhood, particularly among children (the ‘adoption effect’), and about the ways that complete life tables are derived from indirect estimates. Investigation of the estimation procedures suggests that they are very robust for female mortality and acceptably so for male mortality. Small increases in accuracy would accrue from use of a regression based method to estimate male mortality, that incorporates a more sophisticated fertility model than the original method. Such a procedure is presented, together with one for female mortality based on consistent assumptions. Existing methods for fitting life tables to indirect estimates are sometimes less satisfactory. An alternative approach is proposed and assessed. In some countries, orphanhood estimates are seriously biased by the adoption effect. Such errors can be reduced by techniques that analyze data on orphanhood in adulthood. Two such methods are developed and tested. The first estimates mortality from period data on orphanhood after age 20; the second uses data on orphanhood since first marriage. The methods are sensitive to age exaggeration, but data on young adults are a promising source of recent estimates of adult mortality. Finally, procedures are presented for analyzing data on orphanhood prior to marriage. In countries where adults report this information accurately, it can be used to measure adult mortality up to 35 years before the data were collected.