Although analytical procedures for multiple marker risk estimation are now well established, we still lack a unified optimal procedure for deciding which family members to examine and which markers to use. Towards this goal, the application of conditional risk distributions is developed, along with a suggested statistic for judging the utility of a marker. The conditional risk distribution depends on what knowledge has already been obtained about the pedigree, and indicates the expected outcome of risk estimates after another marker is examined. Population genetic aspects including haplotype frequencies, linkage disequilibrium, family size and pedigree structure and the statistical confidence in the linkage map all influence the optimal strategy for multiple marker risk estimation.