For specification of the requirements for efficient cell cryodestruction in tumors, we tested a N2O-driven cryoprobe on experimental models. The cryoprobe was a 3-mm-diameter type for operation via fiber optic bronchoscopes in respiratory medicine. The freezing process, namely the "ice-ball" formation around the cryoprobe tip, was monitored with an impedancemeter. Physical characteristics and formation kinetics of the ice-ball formation (volume, diameter, freezing rate) were studied under defined experimental conditions in various biological liquids, including saline, serum, whole blood, and tumor cell suspensions (rat ascitic hepatoma), either plain or supplemented with gelling agents to approximate solid tumor consistency. Cell destruction (i.e., cryotoxicity to cells) within the ice ball produced in rat ascitic hepatoma was assessed in two ways: the cells, collected after ice-ball thawing, were (1) seeded and cultured according to methods currently in use, or (2) injected into a rat to check for possible development of ascites. Both tests showed that cryotoxicity correlated with freezing rate within the ice ball, cell mortality was total next to the cryoprobe tip (i.e., site of highest freezing rate), while it was absent within the ice-ball periphery. In the area in between, mortality varied gradually. Together our experimental results show that cryotoxicity to cells may be improved by increasing the freezing rate (e.g., by brief precooling of the cryoprobe). Furthermore, for tumor cryosurgery, since cell mortality is maximal next to the cryoprobe, we point out that higher efficacy might be achieved by several overlapping short freezing spots in tumoral tissue, instead of one single prolonged freeze.