Background : The culture of cancer cells has many applications in chemosensi-tivity testing and new drug development. Purpose: Our goal was to adapt simple semiautomated microculture methods for testing the chemosensitiv-ity of melanoma cells freshly recovered from patients' tumors. Methods: Cells were cultured on a substrate of agarose and exposed continuously to cytotoxic drugs, the effects of which were measured by determining the uptake of [ 3 H]thymidine 4–7 days later. Results: Immunocytochemical staining of cells cultured with 5-bromo-2’-deoxyuridine demonstrated that tumor cells were responsible for the measured thymi-dine incorporation. The effects of cytotoxic drugs were calculated as logarithmic 50% inhibitory concentrations and expressed as divergences from the mean in a log-mean graph. The inhibitory effects of amsacrine, etoposide, doxorubicin, cisplatin, mito-mycin C, and fluorouracil were tested. Tumors differed widely in their sensitivity to these drugs, although sensitivity to the three topoisomerase-H-directed agents was highly correlated. Cells from two non-neoplastic hemato-poietic progenitor cell lines (FT and 32D) showed chemosensitivity patterns distinct from those in the melanoma cells, indicating tissue selectivity. Two established melanoma cell lines, MM-96 and FME, were tested under the same conditions and showed sensitivity typical of at least some fresh specimens. Conclusions: These results support the validity of melanoma cell lines as models of freshly resected melanoma cells. If successfully applied to other tumor types, such semiautomated approaches could find wide application in routine hospital laboratories for the chemosensitivity testing of patients‘ tumor cells. [J Natl Cancer Inst 84: 340–345, 1992]