We studied tumor samples from 39 patients, who entered our study from January 1989 to May 1990, to assess whether the ability to establish a continually growing tumor cell line from fresh tumor specimens can be associated with decreased survival times in patients with small-cell lung cancer. The tumor samples were used to establish cell lines in culture using a serum-free medium supplemented with hydro-cortisone, insulin, transferrin, estrogen, and selenium (HITES). Thirty-three of these specimens were obtained by fiberoptic bronchoscopy from primary sites during routine diagnostic procedures. A total of 11 (28%) cell lines were established: seven (21%) from 33 primary tumors and four (80%) from five peripheral lymph nodes. Survival times of the 11 patients whose tumor cell specimens continually grew in culture at any time during their clinical course were significantly shorter than those of the 28 patients whose tumor cell specimens did not grow in vitro (median survival time of 26 weeks versus 73 weeks; P = .0068). Cox's proportional hazards model, including sex, age, Eastern Cooperative Oncology Group performance status, stage, source of specimen, treatment, and in vitro tumor cell growth in the overall patient group, showed that cell line establishment (P = .0017) and no therapy (P = .0015) were the most important factors indicating poor survival time. For the subgroup of 23 primary tumor patients, the important factors (in decreasing order) that indicated decreased order) that indicated decreased survival times were the establishment of a cell line (P = .0112) and treatment with cyclophosphamide-doxorubicin-vincristine alternating with cisplatin-etoposide, versus cisplatin-vincristine-doxorubicin-etoposide therapy (P = .0463). Our study demonstrates that in vitro tumor cell growth is an adverse predominant prognostic factor in patients with small-cell lung cancer. [J Natl Cancer Inst 83:1743–1748, 1991]