Knowledge of the three dimensional dose to normal tissue and tumor is important in both the selection of radiation type and interpretation of clinical results in intraperitoneal radioimmunotherapy. The geometric scale for such calculations ranges from tens of microns for alpha particles to centimeters for beta emitters. In general , the interface dosimetry in intraperitoneal radioimmunotherapy cannot be calculated from image based treatment planning approaches. We have developed a three dimensional calculational model which examines the dose distribution from alpha , beta, and beta / gamma emitting isotopes when an intraperitoneal route of administration is chosen. The model has been used to calculate dose in both idealized geometries and in patient geometries extracted from tissue samples. Consider an irregular peritoneal surface. The dose at a given matrix point is a function of two sources: a) the volumetric distribution from the radiolabeled antibody and b) the surface radioactivity after binding to the tumor. The normal tissue dose distribution is determined by the radioactivity per unit volume of the intraperitoneal fluid and the extent of diffusion of the isotope into the peritoneal tissues. The tumor dose distribution is determined by both the volumetric source and the surface activity due to antibody binding. Dose point kernels for 131-Iodine, 90-Y, 211-At and 212-Pb have been used in the three dimensional Fourier convolution dose calculation. The dose at the fluid / tissue surface is normalized to 100%. The 50% isodose for normal tissue is approximately 1 mm for 90-Y and 20 microns for 211-At with no diffusion into the peritoneal surface. With diffusion to 50 microns, the 50% isodose is at 35 microns for 211-At. The surface activity over tumors is modelled to simulate l%, 10% and 50% binding of radiolabeled antibodies, assuming approximately 10 10 binding sites per cmz. Based on these calculations, the tumor to normal tissue dose ratio is 2O:l for alpha emitters vs 1:l for beta emitters for a 1% binding efficiency. Isodose distributions based on biopsy specimen geometry are presented. The modelling studies suggest that the normal tissue dose may be accurately determined if the solution activity is measured during therapy, and is controllable through adjustment of the solution concentration. The tumor dose can be estimated given the time constant for binding and knowledge of the number of binding sites per unit area. These studies also suggest in-vitro experiments which can be performed to establish relevant dosimetric parameters.