Three adults with acute leukemia were given continuous infusions of 3H‐thymidine for 8 or 10 days. At the end of the infusions 88 to 93% of the leukemic cells were labeled. The leukemic cells varied in size and the flash 3H‐thymidine labeling indexes were generally progressively higher in direct relation to increasing nuclear size. Almost none of the smallest leukemic cells were labeled initially, but, at the end of the infusions, most were labeled regardless of size. The few remaining unlabeled were usually small, and it is concluded that these had remained dormant for die duration of the infusions. The mean generation time of the labeled fraction of leukemic cells in one patient was about 90 hours prior to treatment and about 200 hours during prednisone therapy. In the other 2 patients, antimetabolite therapy was started immediately after the infusions and the generation times were about 100 and 200 hours. There was considerable variability in generation time among the proliferating cells in each leukemic population. The minimum could not be determined because of continual interchange between cells of different sizes as a result of cell growth prior to division and halving of volume thereafter. The data is consistent with the hypothesis that the leukemic cells behave as a self‐maintaining population; we found no evidence for influx of leukemic “stem cells” from an unrecognized precursor compartment. In one patient, destruction of many of the leukemic cells by chemotherapy resulted in an increase in the 3H‐thymidine labeling index and the average size of the remaining cells. Similar findings have been described for several experimental tumors in vivo and for various cell types in vitro; i.e., their‐rate of proliferation varies according to their population density. Acute leukemic cells apparently grow less rapidly when their population density exceeds a critical level, but the finer alterations in kinetics which occur and the mechanism of growth inhibition are as yet unknown.