In the present study a slightly modified MTT assay was used in conjunction with cell counting to determine the antiproliferative efficacy of N-(2-chloroethyl)-N-nitroso-N'-2-hydroxyethylurea (HECNU), vinblastine, and hexadecylphosphocholine (HPC) in a panel of six tumor cell lines. This panel consisted of two human (MDA-MB231, MCF-7) and two rodents (1/C2, 1/C32) mammary-carcinoma cell lines as well as of two tumor cell lines of gastrointestinal origin (HT-29, KB). It was shown that the use of acid isopropanol as a solvent of the formazan crystals produced correlations between cell number and absorption that were as good as, if not better than, those seen after dimethylsulfoxide (DMSO) application. The optimal period of incubation with the MTT dye was 2 h. A comparison of the antiproliferative activity of HECNU revealed that the HT-29 cell line was most resistant [50% inhibition concentration (IC50), 138.7 mumol/l], followed by MCF-7 cells (IC50, 127.7 mumol/l), whereas MDA-MB231 cells showed the highest sensitivity (IC50, 6 mumol/l). Vinblastine induced the highest (MCF-7 cells; IC50, 0.68 nmol/l) and the lowest (1/C2 cells; IC50, 7.69 nmol/l) degrees of growth inhibition in cell lines derived from mammary carcinoma. This contrasted with the activity of HPC, which was considerably less effective in the four mammary-carcinoma cell lines (IC50 from 29.4 to 69.9 mumol/l) than in the two cell lines of gastrointestinal origin (IC50, 1.9 and 3.1 mumol/l). Interestingly, treatment with HPC stimulated the growth of 1/C32 cells in the lower dose range. After treatment with HECNU, the average IC50 value determined in the MTT assay was 2.4-fold that disclosed by cell counting, whereas the average values found for HPC and vinblastine by both methods corresponded fairly well, with the respective values obtained using the MTT assay being only 26% and 14% higher than those measured by cell counting. A dose-dependent increase in the mean size of MCF-7 cells was observed after exposure to HECNU, which--if taken into account--considerably reduced underestimation of this parameter by the MTT assay. No variation in cell size was noted following treatment with HPC and vinblastine. Thus, depending on the antitumor agent used, the MTT assay can result in slight or even considerable underestimation of the antitumor efficacy of certain compounds and may need correction by consideration of the effect of the drugs on cell size.