The in vitro sponge-gel-supported three-dimensional histoculture chemosensitivity assay (Hoffman assay) allows the in vivo-like culture of human tumors. In this study, the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-2H tetrazolium bromide (MTT) end point was applied to the Hoffman assay in an attempt to increase in vitro-in vivo correlation. The chemosensitivities of 16 human tumor lines were determined in vitro by the histoculture assay, and retrospectively correlated to their in vivo chemosensitivity as xenografts in nude mice. The in vitro test was considered to be positive if tumor-cell MTT reduction activity was lowered by more than 50%. The cutoff drug concentrations to determine sensitivity in vitro were determined for mitomycin C, doxorubicin, 5-fluorouracil and cisplatin. Using these cutoff drug concentrations in vitro we found, as a function of time of exposure, a strong correlation between serum drug concentrations found in nude mice given maximum tolerated doses and drug concentrations found in the histoculture media in vitro, thereby establishing a relationship between the amounts of drugs to which tumors were exposed in vivo and in vitro. The overall correlation rate of the efficacy results of the drug-response assay to in vivo chemosensitivities was 89.8%, with 90.0% true-positive and 89.7% true-negative rates, 81.7% sensitivity and 94.6% specificity, thereby indicating potential clinical use for tumor histoculture with the MTT end point.