PURPOSE: There are limited data regarding the role of second-line treatment for metastatic pancreatic cancer (mPC) after the failure of initial chemotherapy. No data exist on the use of GTX after the failure of first-line therapy. PATIENTS AND METHODS: We identified patients who were given GTX chemotherapy for a diagnosis of mPC after the failure of initial therapy. Demographic features, progression-free (PFS) and overall survival (OS), response to treatment, and toxicities were recorded. RESULTS: The 59 evaluable patients received a median of 2 prior therapies. Three had no prior gemcitabine. Median PS was 1. Median survival was 22 weeks; progression-free survival was 9.9 weeks. Survival did not correlate with the number of prior regimens but trended with PS. There were no radiologic responses; those with stable disease (n = 21) had a better survival than those with progression (n = 29) or unevaluable patients (n = 9). Median survival was 38.3, 15.0, and 7.4 weeks, respectively. Grade 3 and 4 toxicities included leucopenia (n = 14), anemia (n = 7), and thrombocytopenia (n = 6). Hospitalizations were required in 21 patients, for febrile neutropenia (n = 7), non-neutropenic infection (n = 3), pulmonary embolus (n = 2), anemia or failure to thrive (n = 9). A 75% drop or more in CA 19-9 correlated with improved survival. CONCLUSIONS: GTX is an active regimen in patients previously treated with gemcitabine for mPC. Better performance status and >75% drop in pretreatment CA 19-9 were associated with longer survival. The number of prior regimens did not predict for survival duration.