The National Cancer Institute (NCI) is implementing a large-scale in vitro drug-screening program that requires a very efficient automated assay of drug effects on tumor cell viability or growth. Many laboratories worldwide have adopted a microculture assay based on metabolic reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT). However, because of certain technical advantages to use of the protein-binding dye sulforhodamine B (SRB) in a large-scale screening application, a detailed comparison of data generated by each type of assay was undertaken. The MTT and SRB assays were each used to test 197 compounds, on simultaneous days, against up to 38 human tumor cell lines representing seven major tumor categories. On subsequent days, 38 compounds were retested with the SRB assay and 25 compounds were retested with the MTT assay. For each of these three comparisons, we tabulated the differences between the two assays in the ratios of test group values to control values (T/C) for cell survival; calculated correlation coefficients for various T/C ratios; and estimated the bivariate distribution of the values for IC50 (concentration of drug resulting in T/C values of 50%, or 50% growth inhibition) for the two assays. The results indicate that under the experimental conditions used and within the limits of the data analyses, the assays perform similarly. Because the SRB assay has practical advantages for large-scale screening, however, it has been adopted for routine use in the NCI in vitro antitumor screen.