Several epidemiologic studies suggest that polychlorinated biphenyl (PCB) levels measured in peripheral blood or adipose tissue are related to increased risk of non-Hodgkin lymphoma (NHL) and, therefore, may be at least partially responsible for the rising incidence of NHL unrelated to HIV infection in recent decades. Case-control studies that measured PCBs in blood, adipose tissue, or household carpet dust, at the time of diagnosis, have observed elevated NHL risk associated with concentrations of either total PCBs or of specific congeners. Similar associations have been found in a number of prospective cohorts. These associations do not seem to be due to confounding by other organochlorines or by other known NHL risk factors. These results support evidence of PCB carcinogenicity from animal studies. However, interpretation of the epidemiologic evidence is limited by the wide range in measurement precision across congeners and by the moderate to high correlation among many congeners. Occupational cohort studies provide very limited support for a relationship between PCBs and NHL. In conclusion, there is mounting evidence of a relationship between certain PCBs and risk of NHL, but important questions remain, especially regarding the magnitude, timing, and causality of that relationship.