We have routinely used a [3H]-uridine microplate assay for assessing chemosensitivity. A colorimetric assay with the advantages of safety, cost and simplicity has previously been described and relies on the ability of living cells to reduce a soluble tetrazolium salt, 3-4,5-dimethylthiazol-2,5-diphenyl-tetrazolium bromide (MMT), into an insoluble formazan precipitate. We compared the chemosensitivity of 14 human tumour cell lines of colonic, lung and cervical carcinoma origin to doxorubicin, vindesine or vindesine immunoconjugates in both the [3H]-uridine assay and a modified MTT assay to evaluate whether we could change to the non-radiolabelled method. Correlation between the concentration of drug causing 50% inhibition of cell growth (IC50) for these agents between the two assays was very poor. However, taking account of recent reports in the literature, we modified the MTT assay by removing serum-containing medium and using dimethyl sulphoxide to solubilise the formazan precipitate. This considerably improved the correlation between the assays for doxorubicin (r = 0.871; P = 0.001) and vindesine (r = 0.981; P less than 0.001). Our data indicates that the MTT assay can be used to replace the [3H]-uridine assay for chemosensitivity screening, but further modifications are necessary to improve the sensitivity and decrease the problem of cell loss after washing, which was noted with some adherent cell lines.