We examined the relationship between prenatal (cord blood) polychlorinated biphenyls (PCBs) and Neonatal Behavioral Assessment Scale (NBAS) performance in babies born to women who consumed contaminated Lake Ontario fish. Cord blood PCBs, DDE, HCB, Mirex, lead, and hair mercury levels were determined for 152 women who reported never consuming Lake Ontario fish and 141 women who reported consuming at least 40 PCB-equivalent lbs. of Lake Ontario fish over their lifetime. Earlier work demonstrated that the newborns of fish eaters are exposed to a more heavily chlorinated distribution of PCB congeners, and that highly chlorinated PCBs (hepta-, octa-, and nonachlorinated biphenyls) are most strongly correlated with breast milk levels, perhaps providing the best index of PCB exposure in the Oswego cohort. Given the above, one would predict that these PCBs would be related to impaired performance on those NBAS clusters associated with fish consumption: namely Habituation, Autonomic, and Reflex clusters of the NBAS. Excepting the Relex cluster, these predictions were confirmed. Results revealed significant linear relationships between the most heavily chlorinated PCBs and performance impairments on the Habituation and Autonomic clusters of the NBAS at 25-48 h after birth. Additionally, higher prenatal PCB exposure was associated with a nonspecific performance impairment on the NBAS as evidenced by a significantly greater proportion of NBAS scales in which poor performance was exhibited (more than 1 standard deviation below the mean) in the most highly exposed neonates. Moreover, PCBs of lighter chlorination were unrelated to NBAS performance, as were DDE, Mirex, HCB, lead, and mercury. These results corroborate our earlier findings linking Lake Ontario fish consumption to the most heavily chlorinated PCB congeners, and suggest that the chlorination and persistence of PCBs may be an important factor both for exposure assessment and for determining relationships with neurobehavioral functions.