The seven cohort studies of the relation between prenatal and postnatal exposure to polychlorinated biphenyls (PCBs) to cognitive, neuropsychological and behavioral development have suggested that exposure to PCBs can cause persistent changes in cognitive functioning. D.V. Cicchetti, A.S. Kaufman, and S.S. Sparrow (this issue) apply six scientific criteria to the design, methodology, and data analysis of the seven cohort studies and demonstrate that numerous flaws limit the reliability and validity of the results and the subsequent conclusions that can be drawn. Their analysis highlights the following general issues for both clinical and research neuropsychologists: (a) self-report may not be the most reliable indicator of exposure to a toxin or of history; (b) statistical results are limited by the appropriateness and reliability and validity of test instruments; (c) clinical examiners must be appropriately trained; (d) some significant findings occur by chance alone, and; (e) the error of overinterpretation of findings--statistical significance does not mean clinical significance and population statistics cannot be applied directly to individuals.