The feasibility of tumor sampling followed by ex vivo assessment of drug sensitivity, using the short-term fluorometric microculture cytotoxicty assay (FMCA), for selection of chemotherapy was investigated prospectively in patients with advanced cancer not amenable to standard treatment. Taxol (175 mg/m2 every 3 wk) was given to patients with tumor samples being low drug resistant (LDR) to Taxol ex vivo, to patients with no LDR drug, and if other drugs were unsuitable. The remaining patients received the most optimal drug(s) based on the FMCA results. Gastrointestinal cancer was dominating among the 61 eligible patients. Tumor sampling was safely performed in 75% by ultrasound-guided core biopsy. Eighty-two percent of the patients had Taxol. Five patients (8%) had a partial remission and 18 (30%) had stable disease. Tumor response was poorly predicted, probably because the Taxol excipient Cremophor EL is cytotoxic exclusively ex vivo. However, patients with tumor cells being LDR to at least one drug ex vivo lived significantly longer than those with no such drug.