The recent publication of many randomised trials about (neo)adjuvant chemotherapy in resectable non-small cell lung cancer (NSCLC) has prompted our group to update a prior meta-analysis of the literature. Randomised studies published in French and English between 1965 and June 2004 were included in this analysis. A qualitative assessment of each trial was first performed using the European lung cancer working party (ELCWP) and the Chalmers' scales. In absence of statistically significant quality difference between positive and negative trials, a quantitative aggregation (meta-analysis) of the individual results was performed. Two trials for which data were available on ASCO virtual meeting website were also included in the meta-analysis. Twenty-five studies eligible for this analysis assessed chemotherapy as induction (n = 6) or adjuvant to surgery (n = 19). No quality difference was detected between positive and negative trials according to the two scores, whatever all trials were combined or only adjuvant chemotherapy studies were considered. The overall meta-analysis showed that the hazard ratio (HR) of the combined results was 0.66 (95% CI 0.48-0.93) in favour of the addition of induction chemotherapy to a standard surgical procedure and 0.84 (95% CI 0.78-0.89) in favour of adjuvant chemotherapy. The effect was significant for adjuvant chemotherapy in stages I and II with a HR of 0.88 (95% CI 0.83-0.94). It was not statistically significant in stage III although the trend was in favour of chemotherapy whatever adjuvant (HR = 0.85; 95% CI 0.69-1.04) or (neo)adjuvant (HR = 0.65; 95% CI 0.41-1.04) chemotherapy was tested. In conclusion, our meta-analysis shows the efficacy of adjuvant chemotherapy in stages I and II resected NSCLC. More data are needed to confirm such a role for induction chemotherapy. Further trials should separate stage III disease from earlier stages.