Concurrent chemoradiotherapy (CCR) followed or preceded by full-dose chemotherapy seems to be a standard treatment for unresectable non-small cell lung cancer (NSCLC). Gemcitabine is a strong radiosensitizer, and a phase I study confirmed the feasibility of CCR with low-dose gemcitabine administered twice-weekly in NSCLC patients. Consequently, we designed a prospective, multicentric, phase II trial to evaluate the efficacy and toxicity of this approach, following induction chemotherapy with cisplatin and gemcitabine. We included patients with unresectable stage III NSCLC, no pleural effusion, adequate pulmonary, renal, liver and hematological functions, Karnofsky index >70 and planned treated volume (PTV) <2200cm3. Treatment consisted of 3 cycles of cisplatin (100mg/m2, d1) and gemcitabine (1250mg/m2, d1 and 8) q3w, followed by CCR (gemcitabine 50mg/m2 on Mondays and Thursdays and radiotherapy 68.4Gy, 1.8Gyqd). After the inclusion of 22 patients (group A), an unacceptable toxicity was detected. Thus, cisplatin dose was reduced to 70mg/m2, and gemcitabine dose was adjusted to 35mg/m2 during CCR. Another 34 patients (33 eligible, group B) were included. Five patients in group A and 6 patients in group B discontinued the study treatment during induction. Thus, 17 and 27 patients, respectively initiated CCR. Hematological toxicity (grades III and IV) was particularly relevant in group A during this phase, with 35 and 23% of thrombopenia and neutropenia, respectively. Nonhematological grades III-IV toxicity of chemoradiation was significant and similar in groups A and B: esophagitis 35.2 and 33.3% and pneumonitis 23.5 and 25.9%, respectively. 40.9% of patients in group A vs. 57.5% in group B completed treatment. Overall response (intention-to-treat analysis) was 68.1% in group A and 63.5% in group B. Median survival was 17.7 months for the whole group with a mean follow-up of 41.2 months. 20% of patients were alive at 3 years. Long-term results of this schedule are encouraging. However, nonhematological toxicity of chemoradiation is substantial and different strategies should be tested to minimize it.