Addition of chemotherapy to the treatment of non-small-cell lung cancer (NSCLC) resulted in a modest but clear improvement in the survival of selected patients. To ascertain if this translates to improved survival in the whole population of patients, we conducted a retrospective population-based study of a sample of 1738 patients diagnosed with primary NSCLC in a French department between 1982 and 1997. The proportion of women, metastatic cases and adenocarcinoma changed significantly over time, as did their management: use of chemotherapy alone increased from 9.7 to 28.1% (P<0.0001), while the use of radiotherapy alone decreased from 32.2 to 9.4% (P<0.0001). The 5-year survival probability was 15.7 % for all patients and 32.6% for those with resectable disease. The 1- and 2-year survival probabilities were 38.2 and 15.6% in locally advanced disease, and were, respectively, 16.8 and 5.2% in metastatic disease. Disease extent and histological subtype were significant independent prognostic factors. Survival of resectable disease was longer among patients treated with surgery or surgery plus chemotherapy, while better outcomes for locally advanced disease were associated with radiation plus chemotherapy. In metastastic disease, patients treated by classical agent without platin or palliative care only had the shortest survival. Despite changes in treatment in accordance with the state-of-the-art, overall survival did not improve over time. It is not unlikely that more patients with bad PS were diagnosed during the latter end of the study period. This could at least partially explain the absence of detection of an overall improvement in survival.