A tumor colony assay (TCA) was evaluated for its ability to predict anticancer drug responses in an N-[4-(5-nitro-2-furyl)-2-thiazolyl]-formamide (FANFT)-induced murine bladder tumor (MBT) model. Ten anticancer drugs were evaluated in vivo and in vitro using four MBT cell lines (40 drug responses). Using the optimum criteria for drug response, the TCA accurately predicted drug responses in the murine model 65% of the time with a true-positive predictive rate of 54% and a true-negative predictive rate of 70%. Drug testing tumor cells immediately on removal from the mouse resulted in a true-positive predictive rate of 77% and a true-negative predictive rate of 100%. The authors conclude that the TCA can predict response to chemotherapy in the MBT model and may be useful in screening investigation compounds for the subsequent evaluation in this murine bladder tumor model.