Background: Advanced pancreatic cancer, even when treated, is highly lethal. The best choice of gemcitabine-based therapy and the prognostic factors affecting the success of treatment remain uncertain. Methods: We identified 159 of 1,475 patients with pancreatic cancer diagnosed in our institution and receiving gemcitabine-based chemotherapy between January 1995 and June 2007. Univariate and multivariate analyses were used to evaluate the prognostic significance of various clinical parameters for overall survival (OS). Results: The median survival after gemcitabine-based therapy was 5.4 months; 89.9% (n = 143) had ductal adenocarcinoma, 55.3% (n = 88) with stage IV. Gemcitabine alone was given to 60 (38%) patients, and gemcitabine with high-dose infusional fluorouracil (5-FU) with (n = 25) or without (n = 39) oxaliplatin was given to 64 (40%) patients. All regimens correlated with OS (p = 0.042) but not with the response rate (RR; p = 0.3). The overall RR was 11.1%, and all responders had a good performance status (PS). The RRs to gemcitabine with infusional 5-FU, and gemcitabine with oxaliplatin and infusional 5-FU were 5.3 and 20.8%, respectively. In a multivariate analysis, old age, advanced stage, poor PS and no enrollment in clinical trials were associated with inferior survival. Conclusions: The outcome for patients who did not participate in clinical trials, regardless of gemcitabine-based treatment, is still bleak.