The MTT assay in agarose, a simple colorimetric test performed in double-layer agarose, has been evaluated for chemosensitivity testing of fresh tumor samples from human cancers in comparison with the MTT assay. The absorbance of cells from fibroblast cell lines or normal tissues was markedly reduced in agarose. The chemosensitivity of cells from a carcinoma cell line or fresh tumor tissues was not apparently affected by the presence of almost 50% of fibroblast or nonmalignant cells in the MTT assay in agarose, whereas it did so in the MTT assay. The frequency of the differences between chemosensitivity of fresh tumor samples in both assays was increased, when the tumor tissue cells contained a higher proportion of nonmalignant cells, i.e. vimentin-positive cells. In 173 patients with various carcinomas, in vitro sensitivity to 7 drugs in the MTT assay in agarose was significantly greater than that in the MTT assay. Further, the MTT assay in agarose had a higher accuracy for prediction of either sensitivity or resistance than the MTT assay in a total of 38 in vitro-in vivo correlations. These results indicated that the MTT assay in agarose was a more suitable technique for the application to chemosensitivity of fresh tumor samples from patients with various carcinomas as compared to the MTT assay.