An attempt has been made to construct an assay of anti-cancer drug sensitivity which is suitable for use with primary culture cells from human bone and soft tissue tumors. The ability of cells to incorporate labeled nucleic acid precursors into acid precipitable material was assessed. Hela-S3 cells were employed to avoid inherited variability and heterogenity of primary cultures of human tumors. Labeled nucleic acid precursors were used not to assay the changes of DNA or RNA synthesis but to detect the viability of cells. A logarithm of counts per minute of incorporated labeled precursors is in proportion to a logarithm of viable cell numbers. This relationship was not influenced by the labeled precursor incubation time. Multiplate was used to provide large numbers of replicate cultures without the requirement of a large numbers of cells. Monolayer Hela-S3 cells which were seeded 24 hours earlier were incubated for three hours with various concentrations of drugs. After removal of drugs, cells were cultured for one week. In this period, Hela-S3 cells with no drug treatment became almost confluent and mitoses occurred about four times. Labeled precursors were incubated for three hours, and incorporated 3H-thymidine was counted. A standard curve of incorporated labeled precursor counts and viable cell numbers was drawn for every assay. The density inhibition of labeled nucleic acid precursor incorporation can be checked and connected with the standard curve, and viable cell numbers after drug exposure can be obtained from the standard curve. When percent survival is plotted against drug concentration, a sigmoid curve is obtained if the drug has dose dependent effect and then the 90% lethal dose (LD90) can be determined from the curve. LD90 was used as the index of anti-cancer drug sensitivity. If the drug has time dependent effect, percent survival of maximal inhibition was used as the index of drug sensitivity.