A rodent bladder cancer model that is induced by intravesical instillation of N-methyl-N-nitrosourea (MNU) was characterized. Cohorts of four to five week old female Fisher 344 rats received four biweekly 1.5 mg. doses of intravesical MNU and were sacrificed at various intervals. By week 13 all animals had flat atypia and/or papillary transitional cell tumors, and 67% of the lesions were moderately (grade II) or poorly differentiated (grade III). By week 20, 83% had gross muscle invasive tumors that eventually killed the host. A cohort of 40 MNU treated animals was subsequently treated commencing at week 17 after initiation of MNU with one of three intravesical six week regimens: 1) saline; 2) BCG (Tice strain); or 3) recombinant human tumor necrosis factor (RTNF) plus adriamycin. There was no difference in animal survival or tumor growth in any group of animals commencing therapy at week 17. A second cohort of 107 animals commenced therapy at 13 weeks after initiation of MNU with one of five intravesical six week regimens: 1) intravesical BCG (Tice strain); 2) adriamycin; 3) recombinant human tumor necrosis factor (RTNF); 4) RTNF plus adriamycin; or 5) BCG plus adriamycin. BCG, RTNF or adriamycin alone had no effect on tumor growth; however, BCG plus adriamycin and RTNF plus adriamycin commencing at week 13 significantly inhibited tumor growth and progression. In conclusion, this autochthonous intravesical rodent transitional cell carcinoma model appears useful for the following reasons: 1) it closely resembles human transitional cell carcinoma histologically and biologically in that all animals develop neoplastic changes in-situ that progress to muscle invasion and kill the host; 2) as with human bladder cancer these tumors do not respond to intravesical therapy if treated when tumor burden is large; however, tumor growth is inhibited when treated early; and 3) this model appears appropriate for screening and developing new intravesical treatments for superficial bladder cancer.