Selection of patients with lung cancer who are most likely to benefit from chemotherapeutic treatment would be a substantial step forward. Therefore, a prospective study in predictive chemosensitivity testing in vitro using the fast green assay (FGA) as developed by Weisenthal et al was carried out. Sixty-six pretreatment tumor specimens were obtained, the majority by means of bronchoscopy (n = 42). Due to an initially insufficient yield of tumor cells (n = 19), dead cells in control samples after four-day culture (n = 15), contamination (n = 7), and laboratory failure (n = 2), only 23 (34.8 percent) samples were successfully tested. In 14 of 36 patients, a comparison between in vitro and in vivo response was possible. Taking into account the number of failures, this number of successful assays does not allow for any conclusion regarding accuracy of the FGA. We conclude that the FGA has limited usefulness for in vitro chemosensitivity testing in patients with disseminated lung cancer in whom biopsy specimens were taken without major surgical investigations. Future directions for predictive testing in vitro are discussed.