Bi-iliac breadth, the frontal maximum diameter between right and left iliac crests, was measured in 1,547 male and 2,085 female residents of a rural area in Japan. All subjects were over 14 years of age. The bi-iliac breadth showed an increase related to age but little sex difference. Modification of age and sex variations from the measured value was obtained by calculation of the score, represented by arithmetical means and standard deviations. Distribution of these scores appeared to be binomial, and since binomial distribution approaches normal distribution when n is large, it is presumed that this trait gains normal distribution. Thus variation of bi-iliac breadth in subjects 20-79 years of age enables us to analyze inheritance. No significant difference was found between husband and wife in the correlation coefficients or between father and daughter (0.13 +/- 0.08). Significant differences were found as follows: father-son (0.35 +/- 0.05) (P less than 0.001), mother-son (0.28 +/- 0.05) (P less than 0.001), and mother-daughter (0.28 +/- 0.06) (P less than 0.001). There was no indication of maternal or paternal effects, since no significant difference was found in father-child and mother-child correlation coefficients. It is concluded that bi-iliac breadth is a quantitative genetic trait under control of polygenes on autosomes. Regression coefficient of child on midparental value was 0.55 +/- 0.05, approximately twice the means of four pairs of correlation coefficients between parent-offspring. Narrowly, heritability was estimated as 0.54 approximately 0.55. Contribution of dominance to total variance was small (VD = 0.11), in contrast to the larger additive genetic variance (VA = 0.54).