The accuracy of extended histopathology to detect immunotoxic chemicals in female B6C3F1 mice was evaluated under the auspices of the National Toxicology Program (NTP). A workgroup was formed consisting of four pathologists who conducted extended histopathological evaluation of lymphoid tissues obtained from a subset of NTP toxicology studies, in which previously detailed immunotoxicity assessment was performed. In addition, a positive control data set of three known immunosuppressive agents, one negative control data set, and an additional negative control group composed of the vehicle only treated groups were included. Data obtained from extended histopathology evaluations were compared to more traditional immune test results (both functional and nonfunctional) from previously conducted immunotoxicity assessments. Analyses of the data indicated that the ability to identify immunotoxic chemicals using histological endpoints decreased linearly as the level of stringency used to determine significant histopathological changes increased. A relatively high (80%) accuracy level was achieved when histological changes were considered in toto (i.e., any histological abnormality in the three tissues examined), using minimal or mild criteria for scoring. When minimal or mild histological changes were considered significant for a specific tissue, a 60% level of accuracy in identifying immunotoxic chemicals was obtained as compared to a 90% accuracy level that was achieved with this data set using the antibody plaque forming cell response, considered to represent the most predictive functional test. A minimal classification was obtained in the analyses of the negative control groups, suggesting that use of the minimal classification for hazard identification is inappropriate as it will likely result in a high incidence of false positives. This was not the case when mild classifications were used as an indicator of significance, which in most instances allowed the successful identification of negatives. When moderate to marked histopathological changes were used to identify immunotoxic chemicals, the level of accuracy that could be achieved was poor. A considerably higher level of accuracy was obtained for the positive control data set than the test chemical data set suggesting that the ability to detect an immunotoxic agent histologically is proportional to the potency of the immunotoxic agent. Comparison of immune function test results and histopathological results obtained from the high-dose treatment groups and the lower-dose treatment group did not reveal any significant differences between the two endpoints to predict immunotoxicity as a function of dose. Of the three lymphoid organs examined, (i.e., lymph node, thymus, and spleen), the most consistent and discernible histological lesions were observed in the thymus cortical region. These lesions correlated with thymus: body weight ratios and to a slightly lesser extent, the antibody plaque forming cell response. Addition of general toxicological endpoints such as body weight and leukocyte counts did not significantly improve the sensitivity of extended histopathology for this data set. Taken together, these data suggest that, while not as sensitive as functional analyses, extended histopathology may provide a reasonable level of accuracy as a screening test to identify immunotoxic chemicals, provided the level of stringency used to score histological lesions is carefully considered to allow for detection of immunotoxic agents while limiting false positives.