In 1999 the Agency for Toxic Substances and Disease Registry (ATSDR) released a Draft Toxicological Profile for Polychlorinated Biphenyls (PCBs). In reviewing the potential human carcinogenicity of PCBs, ATSDR (1999) concluded that "The weight of evidence does not support a causal association for PCBs and human cancer at this time." Just 1 year later, in an updated Toxicological Profile for Polychlorinated Biphenyls (PCBs), the conclusions of another analysis (ATSDR, 2000) on whether exposure to PCBs might represent a carcinogenic risk to humans had dramatically changed to "Overall, the human studies provide some evidence that PCBs are carcinogenic" and "some of these studies provide meaningful evidence that PCBs are carcinogenic in humans." Because this is a substantially different conclusion than that reached only one year previously, it raises a number of questions that must be considered particularly since "weight of evidence" has a precise meaning in the context of evaluating a body of epidemiological data. The present review addresses the additional scientific data that became available between the ATSDR 1999 and 2000 evaluations that was of a magnitude to offset the weight of evidence from numerous epidemiological studies that exposure to PCBs was not causally associated with human cancer to a conclusion only 1 year later that there was now "meaningful evidence" that PCBs posed a carcinogenic risk to humans. Also of interest are the criteria upon which this conclusion is based and the distinction between "weight of evidence" and the newer descriptors of "some evidence" and "meaningful evidence." However, as shown in this review, only one relevant study was published between the ATSDR 1999 and 2000 evaluations and the results of this study were unequivocally supportive of the 1999 conclusion. Because of the continuing controversy surrounding this issue, in this review, all relevant epidemiological data on PCBs are summarized and subjected to another weight of evidence evaluation. This critical review is based on the most recent guidelines (U.S. EPA, 1999a, 2003) for conducting weight-of-evidence evaluations on a body of epidemiological data. Applying a weight-of-evidence evaluation to the PCB epidemiological studies can only lead to the conclusion that there is no causal relationship between PCB exposure and any form of cancer, thereby confirming the conclusions of ATSDR (1999). Also considered is the methodology and logic used by ATSDR (2000) that resulted in overturning the weight of evidence conclusions concerning the human carcinogenicity of PCBs in ATSDR (1999). This issue may have public health and policy implications. It seems appropriate that unbiased evaluations of a body of data, even of controversial issues such as the potential human carcinogenicity of PCBs, be conducted in a transparent manner following applicable guidelines. The dramatic differences between the conclusions of ATSDR (1999) and ATSDR (2000) do not appear to be consistent with this process.