In a series of consecutive 51chromium (51Cr) assays, 21 melanoma and 14 colo-rectal carcinoma patients were tested for their in vitro cell-mediated immune reactions to melanoma, rectal adenocarcinoma, and normal fibroblast target cells. Blood lymphocytes (BL) from four individuals were included in each experiment. In 9/14 experiments the BL effects of 2 melanoma patients were compared to BL effects of 2 colorectal carcinoma patients. In two experiments, BL from 1 melanoma patient and 1 colorectal carcinoma patient were compared for cytotoxic effect with each other and with BL from 2 normal healthy donors, and in the remaining 3 experiments BL from 2 melanoma patients were compared with BL from 1 healthy donor and one patient bearing a tumor which was neither a melanoma nor a colorectal carcinoma. Eleven out of 14 experiments were performed in a criss-cross manner. Target cells in the first three tests of this series consisted of tumor cells and fibroblasts explanted from a single melanoma patient. In all of the remaining experiments, each BL population was tested for cytotoxicity against both a tumor-fibroblast target cell pair explanted from one of two melanoma patients and a tumor-fibroblast target cell pair explanted from a rectal adenocarcinoma patient. Out of 35 tumor patients, 27 (77%) demonstrated a selective cytotoxic effect on the tumor target cells compared to the corresponding fibroblasts, while 4 patients (11%) showed a selective effect on the fibroblasts, and 6/29 patients showed a selective effect on the other type of tumor cells compared to matching fibroblasts. In 8/11 experiments (including two repeat tests) tumor-specific BL effects were demonstrated in a criss-cross manner. BL separations and 51Cr tests were repeated in 11 of the 35 patients 2-4 weeks after their original tests. BL populations from these 11 patients reproduced their individual earlier effects, whether these effects showed specific, non-specific, or no cytotoxicity. In each assay, differences in sensitivity between fibroblasts and tumor target cells in matched pairs were analyzed by comparing the effects of BL from the two controls. No differences in target-cell sensitivity could be demonstrated.