OBJECTIVE Due to the cytotoxicity of DNA-bound iodine-125, 5-[125I]Iodo-2'-deoxyuridine ([125I]IUdR), an analog of thymidine, has long been recognized as possessing therapeutic potential. In this work, the feasibility and potential effectiveness of hepatic artery infusion of [125I]IUdR is examined. METHODS A mathematical model has been developed that simulates tumor growth and response to [125I]IUdR treatment. The model is used to examine the efficacy and potential toxicity of prolonged infusion therapy. Treatment of kinetically homogeneous tumors with potential doubling times of either 4, 5, or 6 days is simulated. Assuming uniformly distributed activity, absorbed dose estimates to the red marrow, liver and whole-body are calculated to assess the potential toxicity of treatment. RESULTS Nine to 10 logs of tumor-cell kill over a 7- to 20-day period are predicted by the various simulations examined. The most slowly proliferating tumor was also the most difficult to eradicate. During the infusion time, tumor-cell loss consisted of two components: A plateau phase, beginning at the start of infusion and ending once the infusion time exceeded the potential doubling time of the tumor; and a rapid cell-reduction phase that was close to log-linear. Beyond the plateau phase, treatment efficacy was highly sensitive to tumor activity concentration. CONCLUSIONS Model predictions suggest that [125I]IUdR will be highly dependent upon the potential doubling time of the tumor. Significant tumor cell kill will require infusion durations that exceed the longest potential doubling time in the tumor-cell population.