Specific congeners of PCBs may differ with respect to their human health risks. For epidemiologic studies, however, measuring levels of specific congeners--as compared with estimating the concentration of total PCBs present, may be of limited value if levels of specific congeners are highly correlated. We examined the correlations among levels of specific congeners in three groups: controls from a case-control study of breast cancer in North Carolina and two groups from Wisconsin with exposure to fish from contaminated waters. Levels of specific congeners were, in general, highly correlated (Pearson r > 0.80). However, the level of congener 180, a heptachlorobiphenyl, tended to be less correlated with levels of lower-chlorinated biphenyls. Among the implications of these findings are that measurement of a select group of congeners may yield essentially the same information as measurement of a large panel, and may be more cost efficient.