Contradictory theories of the mode of inheritance of Perthes' disease have been largely due to sampling error, which is revealed by comparison of four series. Confusion also has been caused by the unequal sex incidence of the disease. Family data from a series of 87 boys and 58 girls with Perthes' disease were combined with those of Gray et al. (223 boys and 44 girls) to provide the largest possible group for analysis. Proportions of first-, second-, and third-degree relatives affected were recorded separately for each sex of index cases and each sex of relatives. Comparison of the incidence of Perthes' disease in relatives with that in the general population of the same sex revealed features of multifactorial inheritance, and a gradient of 35:4:4:1 from first:second:third-degree relatives to the general population. Equal estimates of heritability (h2) from first cousin and sib data suggest that environmental factors are of relatively little importance in etiology. Heritability from sib data was 84 +/- 4%, giving a recurrence risk of 2.6% for sibs and offspring, which agrees with the empirical figure of 2.4%. It is argued that the concentration of cases in certain families is not inconsistent with multifactorial inheritance.