A tetrazolium compound, XTT, bioreducible to a water-soluble formazan was used to develop a simplified cellular cytotoxicity assay. Most (13/15 melanoma and 2/3 colon carcinoma cell lines tested metabolized XTT greater than 50 times more efficiently than the lymphoid effector cells, and thus the test could be performed without separation of the effector from the target cells. The XTT assay (XTT-A) was compared to the standard 51chromium-release assay (51CrA) in terms of sensitivity as well as intra- and interassay variability using low effector to target cell (E:T) ratios and both short and long incubation periods. The correlation coefficient (r) for percent specific lysis (%SL: 35.0 +/- 15.0 versus 30.2 +/- 15.8) or lytic units (LU20/10(7) effector cells: 405 +/- 208 versus 357 +/- 227) between XTT-A and 51CrA was 0.86 for 4 h XTT-A and 51CrA (n = 37). Due to a poor performance of the 51CrA after 24 h incubation of effector and target cells, the correlation coefficient for 24 h assays was reduced to 0.79 (n = 44,%SL = 63.3 +/- 23.9 versus 55.5 +/- 26.6, and LU = 1267 +/- 982 versus 1017 +/- 691). Inter- and intra-assay variability of XTT-A were significantly lower than those for 51CrA. The total background values for XTT-A and 51CrA were similar in 4 h cytotoxicity assay and lower for XTT-A in assays with 24 h incubation. The sensitivity, in terms of discrimination between effector cells with different lytic capacity and targets with different susceptibility, was identical. The XTT-A was simpler, cheaper, and safer to perform than the 51CrA. Furthermore, the XTT-A was suitable for long-term assays and allowed experiments without requiring trypsinization of tumor cells grown in 96-well plates prior to testing.