Histological analysis remains the primary method of distinguishing between small cell (SCLC) and non-small cell lung cancer (NSCLC). This distinction has significant impact therapeutically because of their relative difference in chemoresponsiveness (J.D. Minna et al., Principles and Practice of Oncology, pp. 396-474, 1981). Yet for at least 10% of lung tumors, pathologists will disagree upon the classification (A.R. Feinstein et al., Am. Rev. Respir. Dis., 101: 671-684, 1970). Furthermore, current neuroendocrine markers lack specificity for SCLC although the presence of these markers may help predict chemosensitivity (S.L. Graziano et al., J. Clin. Oncol., 7: 1375-1376, 1989; S.B. Baylin, J. Clin. Oncol., 7: 1375-1376, 1989; C.L. Berger et al., J. Clin. Endocrinol. Metab., 53: 422-429, 1981; A.F. Gazdar et al., Cancer Res., 45: 2924-2930, 1985). In vitro growth characteristics may more accurately reflect biological properties of aggressiveness and susceptibility to chemotherapy. In this study, 3-dimensional gel-histoculture was used to retrospectively distinguish between NSCLC and SCLC. Tumor explants from 78 patients with NSCLC and 13 patients with SCLC were grown in gel-supported histocultures with an overall success rate of 92%. These 2 tumor types were distinguishable by their 3-dimensional in vitro tissue architecture. In addition, proliferation rates were measured by histological autoradiography after 4-day incorporation of [3H]dThd. The percentage of cells labeled in the most proliferatively active regions of the autoradiograms was termed the growth fraction index (A.F. Gazdar et al., Cancer Res., 45: 2924-2930, 1985; R.A. Vescio et al., Proc. Natl. Acad. Sci. U.S.A., 84: 5029-5033, 1987; R.M. Hoffman et al., Proc. Natl. Acad. Sci. U.S.A., 86: 2013-2017, 1989). The mean growth fraction index for pure small cell lung cancer was 79 +/- 10%, differing markedly from that of 35 +/- 19% for mixed small cell/large cell tumors, adenocarcinoma (38 +/- 16%), large cell undifferentiated carcinoma (40 +/- 18%), and squamous cell carcinoma (33 +/- 15%) (P less than 0.001 in each case). We therefore conclude that 3-dimensional gel-histoculture is a useful means of distinguishing pure SCLC from NSCLC, which may improve treatment decision making.