The aim was to make a Monte Carlo simulation approach to estimate the distribution of tumor sizes and to study the curative potential of three candidate radionuclides for radionuclide therapy: the high-energy electron emitter 90Y, the medium-energy electron emitter 177Lu and the low-energy electron emitter 103mRh. A patient with hepatocellular carcinoma with recently published serial CT data on tumor growth in the liver was used. From these data the growth of the primary tumor, and the metastatis formation rate, were estimated. Assuming the same tumor growth of the primary and all metastases and the same metastatis formation rate from both primary and metastases the metastatic size distribution was simulated for various time points. Tumor cure of the metastatic size distribution was simulated for uniform activity distribution of three radionuclides; the high-energy electron emitter 90Y, the mean-energy electron emitter 177Lu and the low-energy electron emitter 103mRh. The simulation of a tumor cure was performed for various time points and tumor-to-normal tissue activity concentrations, TNC. It was demonstrated that it is important to start therapy as early as possible after diagnosis. It was of crucial importance to use an optimal radionuclide for therapy. These simulations demonstrated that 90Y was not suitable for systemic radionuclide therapy, due to the low absorbed fraction of the emitted electrons in small tumors (< 1 mg). If TNC was low 103mRh was slightly better than 177Lu. For high TNC values low-energy electron emitters, e.g., 103mRh was the best choice for tumor cure. However, the short half-life of 103mRh (56 min) might not be optimal for therapy. Therefore, other low-energy electron emitters, or alpha emitters, should be considered for systemic targeted therapy.