Some data suggest that survival in non small cell lung cancer (NSCLC) has improved in the last ten years. A comparison between large prospective therapeutic trials from 1979 to 2002 does show some progress but there are many biases, especially the Will Rogers effect, where stage migration due to new methods of diagnostic imaging and invasive staging procedures could improve actuarial survival in each stage. Prospective randomised trials comparing older and more recent treatments demonstrate some improvements, specifically perioperative chemotherapy in stages I-IIIA, chemo-radiation in unresectable stage III disease and new drugs in stage IV. However retrospective whole population studies fail to show a significant improvement in survival. These discrepancies could be explained by differences between whole populations and those selected for clinical trials. Moreover, the results of the therapeutic trials are not always applied in all unselected patients. In conclusion, NSCLC survival can be improved in a selected population. To reproduce this in the whole population all patients should be managed in accordance with the results of recent trials.