In the recent decade uniform treatment guidelines for non-small cell lung cancer (NSCLC) have been introduced in Sweden. The objective of this study was to examine time trends and differences in treatment intensity for NSCLC between county clinical centres in Central Sweden. A second aim was to investigate whether any differences in treatment of NSCLC were associated with differences in survival. 4345 patients with a diagnosis of NSCLC between 1995 and 2003 were identified in the population-based Lung Cancer Register of Central Sweden. Multivariate logistic regression was used to estimate odds ratios to analyse the likelihood of receiving different treatment modalities for NSCLC. Cox proportional hazard models estimating relative hazard ratios were used to identify factors related to death (by any cause). Of all patients, 33.4% received no treatment, and 17.5% underwent surgery. Between 1995 and 2003, the proportion of patients receiving chemotherapy rose from 14.6% to 55%. There were pronounced differences between county centres in treatment policies, especially concerning surgery and radiotherapy. The likelihood of receiving treatment for NSCLC was highest at county centre A where both surgical treatment and chemotherapy were given more often. Compared to this reference county, the risk of death was between 20% and 40% higher in the other counties after adjusting for age, stage, gender, time period, smoking status and histopathological type. When analyses were adjusted for treatment, county of residence was no longer a prognostic factor. Despite common guidelines there were marked differences in treatment activity between the counties. Treatment activity was associated with survival. Survival benefits may follow improvement in compliance to guidelines.