A number of epidemiological studies have shown predictive relationships between prenatal exposure to polychlorinated biphenyls (PCBs) and subtle deficits in cognitive development in infancy through the preschool years [Child Dev. 56 (1985) 853; J. Pediatr. 116 (1990) 38; J. Pediatr. 134 (1999) 33; Toxicol. Lett. 102-103 (1998) 423; Neurotox. 21 (6) (2000) 1029-1038]. However, since not all studies have demonstrated these relationships (J. Pediatr. 119 (1991) 58-63), debate regarding the role of prenatal PCB exposure in cognitive development continues. The current study was designed to provide additional data to assist in resolving this question. Two hundred twelve children enrolled in the Oswego Newborn and Infant Development Project were assessed using the McCarthy Scales of Children's Abilities at 38 months of age, followed by a reassessment at 54 months of age. The relationship between prenatal exposure to PCBs (cord blood PCBs) and McCarthy performance was assessed at both ages after first controlling for a wide range of important predictors of cognitive development, including socioeconomic status (SES), maternal IQ, maternal education, home environment, cigarette smoking, and many others. Cord blood PCBs were statistically significant predictors of small but measurable deficits in McCarthy performance at 38 months of age. Moreover, a significant interaction between cord blood PCBs and maternal hair mercury (MeHg) was found, such that negative associations between prenatal MeHg exposure and McCarthy performance were found in subjects with higher levels of prenatal PCB exposure. No relationship between PCBs and/or MeHg and McCarthy performance was observed when the children were reassessed almost 1.5 years later (54 months of age). Inspection of the age-related trajectory of McCarthy performance revealed that the more highly exposed children caught up with the least exposed children by 54 months. Although the current data partially replicate the findings of Jacobson et al., Patandin et al., and Walkowiak et al. [J. Pediatr. 116 (1990) 38; J. Pediatr. 134 (1999) 33; Lancet 358 (2001) 1602], results reported here suggest that functional recovery may occur. Moreover, the interaction between PCB and MeHg cannot be considered conclusive until it has been replicated in subsequent investigations.