Pancreatic cancer is a disease with essentially no effective treatment. To increase the potential for discovering effective treatment, we have developed a new treatment model whereby a human pancreatic cancer line, PANC-4, was orthotopically transplanted to the pancreas of nude mice as histologically intact tumor tissue. The tumor grew with subsequent invasive local tumor growth and liver and peritoneal metastases. The antitumor activity of 5-fluorouracil (5-FU) and mitomycin C (MMC) against PANC-4 was initially determined in the in vitro collagen-sponge-gel supported histoculture drug-response assay with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide end point. Inhibition rates were 5.6% for 5-FU and 39.4% for MMC indicating higher efficacy of MMC than 5-FU against PANC-4. When the antitumor activities of 5-FU and MMC against PANC-4 were determined in vivo using the nude mouse orthotopic transplant treatment model, slight local tumor growth inhibition with equivalent incidence of metastases to the liver and the peritoneum as the control were observed in the mice treated with 5-FU, while those treated with MMC had considerably reduced local tumor growth without liver and peritoneal metastases. Thus the histoculture drug-response assay in combination with the orthotopic transplant metastatic models provides for the first time a paradigm for evaluation of agents which may be effective against not only locally growing human pancreatic cancer but resulting metastases as well.