This paper presents results of a study designed to: 1) test for a sex difference in the relative lengths of the finger bones, including the second-to-fourth digit ratio (2D:4D), using left-hand radiographs taken in young children, 2) test whether sex differences can be explained by sex differences in fetal growth, and 3) test the serial stability of sex differences in relative digit lengths, including 2D:4D. Results are presented from 1,060 subjects of the California Child Health and Development Studies. One serial replication at about 9 years old is available from 271 subjects. Results indicate that relative digit lengths are sex-dimorphic in children (Manning et al.  Hum. Reprod. 13:3000–3004,  Early Hum. Dev. 80:161–168). Sex differences in digit length ratios are more pronounced within sibships, where shared family factors are controlled, and are not strongly associated with gross measures of fetal growth, like birth length or weight. Thus, sex differences in the fetal growth of the body are not implicated in sex differences in digital formulae, leaving open the possibility of more direct hormonal and/or genetic causation. However, 2D:4D declined between ages 6–8 in a longitudinal sample, and was a less consistent sex-dimorphic marker than 3D:4D across ethnic groups, suggesting that 3D:4D may be a better marker of perinatal sex differentiation. Prior conflicting findings about 2D:4D may be partly explained by variations in age and ethnicity of populations studied. Am J Phys Anthropol, 2006. © 2005 Wiley-Liss, Inc.