In genetic epidemiology, adoption and twin study designs are the most commonly used designs to identify genetic and common environmental components underlying familial aggregation. Other fixed family set designs such as a half-sibling design can also be applied to test the same hypotheses. The feasibility of half-sibling designs for detecting a genetic (single-locus or multifactorial) source of familial aggregation of a disease is studied and is compared to both adoption and twin study designs. Results are presented for two types of design where the sampling units are relative pairs reared apart (I) or reared together (II). Although they generally require more observations to achieve the same power, designs involving half-siblings may be more feasible in situations where there is greater availability of these relatives. © 1994 Wiley-Liss, Inc.