Three separate phase III randomized studies were conducted to compare the efficacy of a gemcitabine plus cisplatin combination (GC) with other chemotherapy regimens in the treatment of European and North American patients with inoperable (stage IIIB or IV) non-small cell lung cancer (NSCLC). The Spanish Lung Cancer Group (SLCG) compared the GC regimen with an etoposide plus cisplatin combination (EP), the Hoosier Oncology Group (HOG) compared it with single-agent cisplatin, and the Italian Lung Cancer Project (ILCP) compared it with a mitomycin plus ifosfamide plus cisplatin combination (MIC). The three studies each had a different primary objective (response rate, survival, or quality of life, respectively). From July 1995 to February 1997, 751 patients were enrolled into the three studies. In the SLCG study, 69 were entered in the GC arm and 64 in the EP arm; in the HOG study, 155 were entered in the GC arm and 154 in the cisplatin-alone arm; and in the ILCP study, 154 were entered in the GC arm and 152 in the MIC arm. In the SLCG study, gemcitabine 1,250 mg/m2 was given by 30-minute infusion on days 1 and 8 of each 21-day cycle. In the HOG and ILCP studies, gemcitabine 1,000 mg/m2 was given on days 1, 8, and 15 of each 28-day cycle. Cisplatin 100 mg/m2 was given on day 1 in the SLCG and HOG studies and on day 2 in the ILCP study. In the EP arm, etoposide 100 mg/m2 was given on days 1, 2, and 3 of each 21-day cycle. In the MIC arm, mitomycin 6 mg/m2 and ifosfamide 3 g/m2 were given on day 1 of each 28-day cycle. In the SLCG study, 28 patients (40.6%) in the GC arm and 14 patients (21.2%) in the EP arm achieved an objective response (P = .01). In the HOG study, 48 patients (31%) in the GC arm and 14 patients (9.1%) in the cisplatin-alone arm attained on objective response (P < .001). In the ILCP study, 61 patients (40%) in the GC arm and 42 patients (28%) in the MIC arm achieved an objective response (P = .03). Progression-free time was significantly longer for the GC arm (6.8 months) than for the cisplatin-alone arm (4.2 months) (P < .001). The median survival time was remarkably identical for the GC arms in the three studies (8.7 months), whereas it was 7.2 months for the EP arm, 7.6 months for the cisplatin-alone arm, and 9.5 months for the MIC arm. The main toxicities were myelosuppression and vomiting. The assessment of quality of life in the ILCP study is ongoing. Both 21- and 28-day cycles of GC were effective in the treatment of NSCLC, and this combination can be considered as a current "standard" therapy for advanced NSCLC patients.