Background: The FAST is a 2 × 2 factorial trial addressing two questions: (1) the role of replacing cisplatin (P) with a non-platinum agent, vinorelbine (N), and (2) the role of adding a third agent, ifosfamide (I), in a doublet based on gemcitabine (G). Methods: A total of 433 stage IIIB–IV non-small cell lung cancer (NSCLC) patients were randomised to one of four arms: gemcitabine–cisplatin (GP), gemcitabine–vinorelbine, gemcitabine–ifosfamide-cisplatin or gemcitabine–ifosfamide–vinorelbine. Two comparisons were performed: N- vs P-containing regimens and I-triplets vs non-I doublets. Results: For N- vs P-containing regimens, adjusted overall survival was 9.7 vs 11.3 months (P=0.044), progression-free survival was 4.9 vs 6.4 months (P=0.020) and response rate was 24% vs 31% (P=0.124), respectively. No statistically significant difference was observed between doublets and triplets. Grade 3–4 haematological toxicity was significantly more frequent in P-containing therapy; grade 3–4 leucopenia was significantly more common in triplets. Concerning non-haematological toxicity, grade 3–4 nausea-vomiting was significantly increased in P-containing regimens. Conclusions: This trial provides evidence of a slight survival superiority of GP-containing regimens over platinum-free N-containing chemotherapy. This trial also confirms that the addition of a third chemotherapy agent (I) to a standard G-based doublet does not improve treatment outcome.