If total tumor cell kill is a requirement for "cure," as it probably is for at least some tumors, treatment of micrometastases containing less than or equal to 10-6 viable tumor cells is a problem facing all cancer therapists. Metastatic tumor cell foci of less than or equal to 10-6 cells, particularly if widely disseminated, are, in the main, grossly undetectable and are, therefore, the proper targets for the chemotherapist and the immunotherapist. Current knowledge of tumor cell population growth kinetics indicates that the growth fraction (viable cells undergoing active cell replication) is inversely related to population size. Micrometastases should, therefore, be more sensitive to antimetabolite (cell-cycle-specific) anticancer drugs than the larger, grossly apparent, primary tumor from which they were derived. In fact, drug response of the primarytumor may not reliably predict the sensitivity of micrometastases to cell-cycle-specific drugs. In addition, optimal drug treatment scheduling for effective, but noncurative, drugs against clinically recognized tumors may not predict optimal scheduling of the same drugs against micrometastases. This is shown particularly in cell-cycle-specific drugs, since the generation time of tumor cells in small population foci (less than 10-7 cells) is often significantly shorter than in larger population foci (greater than 10-7 cells). Since first-order cell kill kinetics characterizes effective drug kill of tumor cells and the best currently known anticancer drugs have variable but generally limited tumor cell kill potential, drug treatment of likely micrometastases should be started as soon after the end of likely noncurative radiologic or surgical treatment as possible. Experimental data indicating that significantly smaller numbers of viable tumor cells can establish lethal tumors in the presence of radiation-inactivated tumor cells than in their absence suggest that small populations of residual viable tumor cells in radiation-treated tumor sites may be a greater threat to clinical cure than similar sized populations remaining in situ after surgery. Experimental data supporting the above concepts are presented and discussed.