Although clinical trials suggest that chemotherapy can improve survival for both resected and unresected pancreatic cancer patients, the extent to which it is used in routine clinical practice is unclear. We conducted a population-based investigation of treatment patterns and factors influencing treatment receipt and mortality for pancreatic cancer. We included 3173 patients with primary invasive pancreatic cancer, diagnosed in 1994-2003, from the National Cancer Registry (Ireland). Analysis was done by joinpoint regression, logistic regression and Cox proportional hazards. Propensity score methods were used to compare mortality in those who received chemotherapy and in 'matched' patients who did not. Seven percent of patients had a resection and 12% received chemotherapy. The resection rate did not change significantly over time and less than a quarter of patients with localised disease underwent resection. Chemotherapy use increased by 20% per annum, reaching 20% among unresected and 39% among resected patients in 2002-2003. Forty two percent of patients were untreated, and this percentage was unchanged over time. After adjusting for clinical factors, patient characteristics were significantly associated with treatment receipt; older and unmarried patients were less likely to be treated. Among resected patients, risk of death fell by 10% per annum. Chemotherapy receipt was associated with significantly reduced mortality among both surgical (hazard ratio (HR)=0.50, 95% confidence intervals (CIs) 0.27-0.91) and non-surgical patients (HR=0.48, 95% CI 0.38-0.61). Our findings suggest that there may be potential for extended dissemination of chemotherapy, and possibly also for greater utilisation of curative resection, in routine practice which, in turn, has potential to improve survival at the population level.