It was recently shown that hand postaxial polydactyly differed from foot postaxial polydactyly. The aim of this work was to test whether thumb and hallux duplication also had different clinical and epidemiological characteristics, depending on limb involvement. We studied 920 newborn infants with first digit duplication, ascertained among 3,444,374 births by the Latin-American Collaborative Study of Congenital Malformations (ECLAMC), from 1967 to 1995. Since biphalangeal thumb duplication or hallux duplication can occur in families with triphalangeal thumb or polysyndactylous propositi, these groups were also analyzed. The 715 isolated (nonsyndromal) cases (prevalence 2.08 per 10,000) were subdivided into five groups: thumb duplication (N = 568; prevalence: 1.65/10,000); hallux duplication (N = 82; prevalence: 0.24); thumb and/or hallux duplication plus syndactyly (polysyndactyly) (N = 37; prevalence: 0.11); triphalangeal thumb (N = 24; prevalence: 0.07), and thumb duplication plus hallux duplication (N = 4; prevalence: 0.01). Both thumb and hallux duplication groups showed a significant excess of males, and right sidedness was also more frequent in both of them, though without statistical significance for hallux duplication. Thumb duplication was more often unilateral (94.7% versus hallux duplication of 81.5%), and its prevalence was higher in Bolivia (3.37/10,000) than in the other 10 Latin-American countries included (1.62/10,000). In a subseries of 405 preaxial polydactylies with matched controls, a logistic regression analysis showed that birth weight and gestational age had an effect on the calculated risk of having an infant with thumb duplication, while first trimester vaginal bleeding had only a borderline effect. None of the polydactyly groups showed abnormal values for twinning, perinatal mortality, ethnicity, maternal education, parental ages, parity, parental subfertility, or consanguinity. There were 70/405 familial cases. Their pedigrees were compatible with autosomal dominant inheritance with a 9% penetrance for thumb duplication and hallux duplication and a 70% penetrance for triphalangeal thumb and polysyndactyly. Inheritance of thumb duplication and probably the untested inheritance of hallux duplication were also compatible with a four-locus multiplicative model. The observed differences in laterality, geographical distribution, birth weight, gestational age, and first trimester vaginal bleeding between thumb duplication and hallux duplication groups suggested that apparent preaxial polydactyly type 1 is a causally heterogeneous group.