We conducted a meta-analysis of the association between leisure time physical activity (LTPA) and risk of pancreatic cancer to update previous analyses in light of newly published studies, to examine subgroups of interest and potential sources of heterogeneity. We searched the PubMed and MEDLINE databases for studies until February 2015. Study information was collected using a standardized form to abstract relevant data on study design, number of cases, participant and study characteristics, assessment of LTPA, risk estimates, and adjustments for confounding by two independent abstractors. We used randomeffects models to pool estimates from included studies of lowest versus highest comparison of LTPA. The search identified 26 studies eligible for inclusion into the meta-analysis. The combined summary risk estimate was [relative risk (RR), 0.89; 95% confidence interval (CI), 0.82-0.96]. There was evidence of heterogeneity across studies (I 2 ¼ 22.1%, P heterogeneity ¼ 0.130). Some of the heterogeneity could be explained by study design, with stronger protective effects observed among casecontrol studies (RR, 0.69; 95% CI, 0.59-0.81) compared with cohort studies (RR, 0.96; 95% CI, 0.91-1.02). Across study designs, age of population was a source of heterogeneity, with stronger effects observed among younger (<50 years) populations. The present meta-analysis supports a protective association between LTPA and pancreatic cancer with an 11% risk reduction observed. LTPA appears to have the strongest effect among young populations. Additional investigations are needed to provide insights regarding the impact of LTPA in healthy adult populations, to reduce the risk of pancreatic cancer and encourage increases in LTPA. Cancer Epidemiol Biomarkers Prev; 24(10);