The paper reviews methods of studying cell kinetics in man, cell population kinetics of human tumors and bone marrow, drug interactions and the cell cycle, and possible applications to chemotherapy. The conclusions drawn are: (1) Cell cycle time and S-phase duration for proliferating granulocyte precursors in human bone marrow are poorly defined but are probably shorter than median values for most human tumors, including leukemia. (2) Most drugs have greater toxicity for cycling cells and some variation in toxicity at different phases of the cell cycle. There is a special need for chemotherapy directed at slowly proliferating and hypoxic tumor cells. (3) Pretreatment indices of tumor cell kinetics are of little value in choosing drugs or in predicting response. (4) Experiments in animals have demonstrated that therapeutic index may depend on schedule. Knowledge of cell kinetics in animals rarely allows prediction of the optimal schedule and is unlikely to do so in man. Optimal schedules in mice are not directly relevant to man. (5) Measurement of tumor labeling index or DNA histogram by flow microfluorimetry to detect cell synchrony is of little benefit in scheduling if concurrent changes in bone marrow are ignored; these methods are invalid at short intervals after treatment because surviving clonogenic cells are indistinguishable from a larger number of drug-damaged cells prior to their lysis. (6) The major factor determining the outcome of chemotherapy is the availability of drugs with activity for the tumor and acceptable host toxicity. Claims that complex schedules using several drugs are effective because of synchrony or kinetic differences of tumor and normal tissue are at present unsubstantiated.