BACKGROUND: Epidemiologic data regarding the potential neurotoxicity of perfluorinated compounds (PFCs) are inconclusive. We investigated the associations between in utero exposure to perfluorooctanoic acid (PFOA) and perfluorooctyl sulfonate (PFOS) and early childhood neurodevelopment. METHODS: We recruited 239 mother-infant pairs in northern Taiwan from the Taiwan Birth Panel Study, which was established in 2004. We examined the association between PFCs in cord blood and children's neurodevelopment at 2 years of age, using the Comprehensive Developmental Inventory for Infants and Toddlers. This tool contains cognitive, language, motor, social, and self-help domains; test scores were further transformed into developmental quotients according to standardized norms. All multivariate regression models were adjusted for infant sex and gestational age, maternal education, family income, cord blood cotinine levels, postnatal environmental tobacco smoke exposure, and breastfeeding. RESULTS: Prenatal PFOS concentrations in both untransformed and natural log (Ln)-transformed values were associated with adverse performance on the whole test and the domains related to development. A dose-response relationship was observed when PFOS levels were categorized into four groups. This association was most obvious in relation to the gross-motor subdomain. Across the PFOS interquartile range, the quotients of the gross-motor subdomain decreased by 3.7 points (95% confidence interval [CI] = -6.0 to -1.5), with an increasing odds ratio of poor performance (2.4; 95% CI = 1.3 to 4.2). In contrast, measures of association between PFOA concentrations and test scores were close to null. CONCLUSIONS: Prenatal exposure to PFOS, but not PFOA, may affect children's development, especially gross-motor development at 2 years of age.