PURPOSE: This randomized phase II trial evaluated two docetaxel-based regimens to see which would be most promising according to overall response rate (ORR) for comparison in a phase III trial with epirubicin-cisplatin-fluorouracil (ECF) as first-line advanced gastric cancer therapy. PATIENTS AND METHODS: Chemotherapy-naïve patients with measurable unresectable and/or metastatic gastric carcinoma, a performance status <or= 1, and adequate hematologic, hepatic, and renal function randomly received <or= eight 3-weekly cycles of ECF (epirubicin 50 mg/m(2) on day 1, cisplatin 60 mg/m(2) on day 1, and fluorouracil [FU] 200 mg/m(2)/d on days 1 to 21), TC (docetaxel initially 85 mg/m(2) on day 1 [later reduced to 75 mg/m(2) as a result of toxicity] and cisplatin 75 mg/m(2) on day 1), or TCF (TC plus FU 300 mg/m(2)/d on days 1 to 14). Study objectives included response (primary), survival, toxicity, and quality of life (QOL). RESULTS: ORR was 25.0% (95% CI, 13% to 41%) for ECF, 18.5% (95% CI, 9% to 34%) for TC, and 36.6% (95% CI, 23% to 53%) for TCF (n = 119). Median overall survival times were 8.3, 11.0, and 10.4 months for ECF, TC, and TCF, respectively. Toxicity was acceptable, with one toxic death (TC arm). Grade 3 or 4 neutropenia occurred in more treatment cycles with docetaxel (TC, 49%; TCF, 57%; ECF, 34%). Global health status/QOL substantially improved with ECF and remained similar to baseline with both docetaxel regimens. CONCLUSION: Time to response and ORR favor TCF over TC for further evaluation, particularly in the neoadjuvant setting. A trend towards increased myelosuppression and infectious complications with TCF versus TC or ECF was observed.