Since the description of human thrombopoietin (TPO) we investigated the thrombocytosis-inducing capacity of human serum samples derived from individuals with altered thrombocytopoiesis. Several times the degree of thrombocytosis developing in recipient mice differed markedly even when applying the same human material. In the last 2 years, we applied single doses of recombinant human TPO (rHuTPO) to random-bred CFLP mice, and the same observation was made. Taken together with previous information (before 1970) it was possible to select cases in which the percent increases in circulating platelet counts inversely correlated with the starting levels. It appears, however, that apart from the known absorbing role of platelets and megakaryocytes, the response to single doses of exogenous rHuTPO in mice depends, at least partially, on an unknown endogenous homeostatic mechanism. Mixing thrombopoietically active human sera with platelet-free normal serum in a 1:1 ratio remarkably reduced the thrombocytosis-inducing capacity. Repeated pharmacological doses of TPO, applied in the majority of the reported trials, however, easily obscure the physiological control mechanism.