Background: Platinum-based chemotherapy (CT) is the standard treatment for patients (pts) with advanced NSCLC, but the evidence of its efficacy among ECOG PS2 pts is weak, because these pts are usually excluded from clinical trials; concern exists about tolerability and feasibility of standard CT in these pts. No prospective randomized trial has tested the addition of cisplatin to single-agent CT in pts with advanced NSCLC and PS2. Methods: CAPPA-2 was a multicentre, randomized phase III study for first-line treatment of PS2 pts with advanced NSCLC. Patients, aged 18-70, were eligible if they had stage IV or IIIB with malignant pleural effusion or metastatic supraclavicular nodes (TNM VI ed.) and adequate organ function. Patients in standard arm received gemcitabine 1,200 mg/m² dd1 and 8.Patients in experimental arm received cispaltin 60 mg/m² d1 plus gemcitabine 1,000 mg/m² dd1 and 8. All treatments were repeated q3w, up to 4 cycles, unless disease progression or unacceptable toxicity. Primary endpoint was overall survival (OS). To have 80% power of detecting hazard ratio (HR) 0.71, corresponding to an increase in median OS from 4.8 to 6.8 months, 285 deaths were required. Results: The study was stopped in June 2012 after the enrolment of 57 pts, due to the slow accrual and the report of positive results from a similar study. Median OS was 3.0 months with single-agent gemcitabine and 5.9 months with cisplatin + gemcitabine (HR 0.52, 95% CI 0.28-0.98, p=0.039). Combination CT produced longer PFS (median 1.7 vs. 3.3 months, HR 0.49, 95% CI 0.27-0.89, p=0.017) and higher response rate (4% vs. 18%, p=0.19), without substantial increase in toxicity. Conclusions: Addition of cisplatin to single-agent gemcitabine improves survival as first-line treatment of PS2 patients with advanced NSCLC. Clinical trial information: NCT00526643.