Background: So far there are few published data on optimal duration of chemotherapy in patients with advanced non small cell lung cancer (NSCLC) We designed a phase III trial to compare 4 cycles vs 6 cycles of chemotherapy using a standard combination. Primary endpoint is overall survival (OS). Secondary endpoints included time to progression (TPP), overall response rate (ORR), one-year survival (1y -S) and safety. Methods: Randomization was stratified by stage (IV vs IIIB) and performance status (0/1 vs 2). Patients received gemcitabine 1250 mg/m 2 , day 1 and 8 + Carboplatin, AUC 5, day 1, every 21 days. Eligibility criteria: age > 18 years, histologically proven NSCLC, weight loss < 10%, no brain metastasis, adequate renal, hematological, hepatic functions and informed consent. The efficacy analysis (OS, TTP, 1y-S) will be performed on intent to treat basis. Results: Between October 2002 and December 2004, 220 pts were enrolled. Arm A -4 cycles (n=110) and Arm B -6 cycles (n= 110) were well-balanced for patient characteristics: median age (A/B): 64,7 vs 63,9 yrs; male/female (A/B) 86/24 vs 86/24; IIIB/IV (A/B) 24/86 vs 29/81; PS 0,1 vs 2 (A/B) 90/20 vs 91/19; adenocarcinoma (A/B) 59 vs 57; squamous (A/B) 35 vs 37. ORR and toxicity was evaluated in 204 pts (7.2% pts were not evaluated). ORR% (43,8 vs 47,3); median number courses (3,5 vs 4,8). Grade 3/4 toxicities included neutropenia (10,2% vs 13,6%); thrombocytopenia (3.1% vs 5,2%); anemia (1,0% vs 2,1%). All grades of nausea/vomiting (14,2% vs 16,8%). Fifteen pts (Arm A) and sixteen pts (Arm B) have had a second line therapy (docetaxel) after disease progression. At Jan/2007, 107 at each arm died. Time to progression was not statistically different between the two treatment arms (p=0.077). Median time to progression was 4 months (95% CI: 3.2-4.8 months) for patients randomized to 4 cycles and 5 months (95% CI: 3.8-6.2 months) for patients randomized to 6 cycles. Overall survival was statistically different between the two treatment groups (p=0.047) and favourable to 6 cycles. Median survival was 7 months (95% CI: 5.9-8.1 months) for 4-cycles regimen and 12 months (95% CI: 9.8-14.2 months) for 6-cycles regimen. When compares patients according to the number of cycles they really received and according to the randomisation regimen assigned: 79 patients in 4 cycles and 65 in 6 cycles. Time to progression and overall survival was statistically different between the two treatment groups (p=0.018) and favourable to the regimen of 6 cycles Conclusions: Overall survival rate was statistically different between 4 vs 6 cycles. TTP, ORR and One year survival was not statistically different. In patients without progression or major toxicity during chemotherapy, we recommend six cycles in first line chemotherapy.