The retrospective evaluation of the dose after an irradiation accident is of paramount importance; it allows an adequate selection of patients and the most appropriate treatment can then be proposed. Classical physical dosimetry often lacks precision for dose assessment in such accidents. Cytogenetics, usually more reliable, is not 100% accurate and cannot be used in some particular instances. At the Institut Gustave-Roussy, we studied amylasemia in 15 patients who received a total body irradiation (TBI) for bone marrow grafting, at various dose levels (10, 2 and 1.35 Gy). Hyperamylasemia was found to be constant and dose-dependent. Ten additional patients given a localized irradiation of 2 Gy in the Waldeyer ring had a similar rise in amylasemia as did TBI patients who had received the same dose. In contrast, 13 patients given a pancreatic irradiation (as part of a localized abdominal irradiation) did not show any increase in amylasemia. This study seems to confirm reported data, which suggested that post-TBI hyperamylasemia is almost only related to salivary gland irradiation. Amylasemia could possibly be used as a "biological dosimeter"; however, the dose-effect relationship should be more precisely defined, as well as individual variations. Moreover, the definition of a "threshold-dose" below which hyperamylasemia can never be detected, would be of interest for radioprotection.