What conclusions can be drawn concerning the potential neurological effects of perinatal exposure to either PCBs, or PCBs and other fish-borne contaminants? First, by their very nature epidemiological studies are limited in their ability to detect subtle associations--including possible links between exposure to low levels of environmental contaminants and disease. As stated by Dr. Schantz, both Rogan and the Jacobsons report small changes in motor and cognitive behavior--typically less than one-half of a standard deviation--and only in the most highly exposed children. Given these small changes in CNS function, the substantive criticisms of Paneth (including the Jacobsons' choice to employ a random, rather than matched, control sample and the related fact that fish-eating mothers differed from non-fish-eating mothers on several important characteristics) and similar "generic" concerns raised by Taubes, a critical reader must question both the validity of the findings from the Michigan study and the reasons for discrepancies in results between the Jacobson and Rogan studies. Are the differences in neurobehavioral effects reported by the Jacobsons and colleagues, and Rogan and colleagues, due to the presence of confounders, exposure to different neurotoxicants, or subtle differences in methodologies? At present it is not possible to answer these questions. Nevertheless, certain commonalities exist between the Rogan and Jacobson studies, and most recently, the study conducted by Daly and colleagues in New York. All of these studies report alterations in the Brazelton Neonatal Behavioral Assessment Scale, suggesting that exposure to environmental contaminants (including PCBs) may induce subtle, transient alterations in maturation of the human CNS. Secondly, because contaminated fish contain a large number of putative developmental neurotoxicants (e.g., methyl-mercury, p,p'-DDE, PCBs, and pesticides), I am pessimistic that additional studies of human populations exposed to contaminated aquatic and marine fish and mammals will allow investigators to determine which contaminant(s) may be responsible for the observed association between fish consumption and neurobehavioral deficits. For example, although PCB body burdens have been measured in the majority of the epidemiological studies, PCBs may simply reflect exposure to other fish-borne contaminants. In light of the above statements, future epidemiological studies should focus on highly exposed susceptible populations such as occupationally exposed workers or the aged. Results from these studies would provide important information on the risk of perinatal or adult exposure to PCBs in susceptible populations, although generalization of results obtained in these populations to the general population may be fraught with difficulties. Finally, because of the limitations of epidemiological studies, particularly those studying fish-eating populations, future risk assessments should depend more heavily on laboratory derived data, including studies in nonhuman primates exposed to environmentally relevant mixtures and relevant doses of PCB congeners and other known or suspected neurotoxicants.