In order to assess the usefulness of chemosensitivity tests in the treatment of colorectal cancer, 71 tumor specimens were tested for chemosensitivity in the following assays: nude mouse isotope assay (NMIA), subrenal capsule assay (SRCA), human tumor clonogenic assay (HTCA) and adenosine triphosphate inhibition assay (ATPA). The agents examined were: mitomycin C (MMC), 5-fluorouracil (5-FU), cyclophosphamide (CPM), adriamycin (ADM) and cis-diamminedichloroplatinum (CDDP). The evaluability rates were 90.8, 93.9 and 92.3 per cent in NMIA, SRCA and ATPA, respectively, but only 42.9 per cent in HTCA. The tumor response rates were 50.8, 45.2, 16.7 and 33.3 per cent in NMIA, SRCA, HTCA and ATPA, respectively. Individual drug sensitivity rates differed among all 4 assays, ranging from 0 to 33.3 per cent. In the arbitrary judgment of the 4 assays, the most sensitive agent was CDDP, followed by CPM, ADM, 5-FU and MMC. In the prospective study, predictive accuracy rates of the clinical responses were 81.3, 66.7, 100, 100 and 76.5 per cent in NMIA, SRCA, HTCA, ATPA and the arbitrary judgment, respectively. A significant correlation between the survival time and the results of SRCA was detected retrospectively. These results suggested that colorectal cancer might not be completely resistant to anticancer agents, and that chemosensitivity tests might be useful in the individual therapy of colorectal cancer patients.