The aim of this study was to evaluate the therapeutic possibilities of radiolabeled antisense oligodeoxynucleotides. The internal radiation dose from known cellular and animal data was calculated, and the suitability of different phosphorus isotopes as labels for oligonucleotides was assessed. We calculated the pharmacokinetics and tissue distribution in vivo of short oligodeoxynucleotide phosphorothioates by using the data from two different radionuclides: phosphorus-33 (t1/2 = 24.4 days, maximum beta-energy = 250 ke V) and phosphorus-32 (t1/2 = 14.3 days, maximum beta-energy = 2270 ke V). The absorbed doses of 32P-labeled and 33P-labeled oligonucleotides were estimated using the published biodistribution data for several oligonucleotides in animal models for both tumor xenografts and AIDS. The local absorption of 33P was higher than that of 32P if the radius of the spherical distribution of activity was smaller than approximately 500 microns. In a mouse tumor xenograft model, an intravenously injected activity of 1 MBq achieved sufficient radiation doses in the tumor; 11 Gy for 32P and 1.5 Gy for 33P were obtained. In normal organs in the same model, the liver doses were 5.0 Gy (32P) and 0.7 Gy (33P), and the kidney doses were 14 Gy (32P) and 2.0 Gy (33P). We conclude that 33P may have more beneficial radiotherapeutic characteristics for oligonucleotides than 32P. This method could be applied on the macroscopic, cellular, and subcellular levels and help to design further experimental studies for the use of oligonucleotide radiotherapy and phosphorothioate probes.