Background: There is a need for prognostic tools in metastatic pancreatic cancer (MPC) because they have the potential to optimize patients’ selection in clinical trail and guide treatment strategies. We address this issue by developing and validating a prognostic nomogram with open clinical trail data. Methods: Data from the comparator arm of three clinical trails (NCT00844649, NCT01124786 and NCT00574275) in MPC treated with gemcitabine as first-line chemotherapy were analyzed. The former two were taken as training cohort while NCT00574275 was used as validation cohort. Cox regression model was used to investigate prognostic factors from twenty-three baseline characteristics in training cohort. Based on these factors, a nomogram was developed and internally validated with Harrell’s Concordance index (C-index) and calibration plots. It was further externally validated in the validation cohort. Results: In the training cohort (n = 445), performance status, liver metastasis, CA19-9 log-value, neutrophil, platelet (PLT) and albumin were independent prognostic factors for overall survival (OS). A nomogram based on these factors was generated to predict OS and survival probabilities. The nomogram showed an acceptable discrimination ability (C-index: 0.683) and good calibration. The prognostic score based on nomogram could stratify patients into three-risk groups with median OS of 11.7, 7.0 and 3.7 months (P < 0.001). The nomogram was further externally validated in validation cohort (n = 273, C-index: 0.699); median OS of 10.6, 7.3 and 4.0 months for the three-risk groups (P < 0.001). Conclusions: The prognostic nomogram can accurately predict OS for MPC treated with gemcitabine as first-line chemotherapy.