Physical activity may prevent pancreatic cancer by regulating body weight and decreasing insulin resistance, DNA damage, and chronic inflammation. Previous meta-analyses found inconsistent evidence for a protective effect of physical activity on pancreatic cancer but those studies did not investigate whether the association between physical activity and pancreatic cancer varies by smoking status, body mass index (BMI), or level of consistency of physical activity over time. To address these issues, we conducted an updated meta-analysis following the PRISMA guidelines among 30 distinct studies with a total of 10,501 pancreatic cancer cases. Random effects meta-analysis of cohort studies revealed a weak, statistically significant reduction in pancreatic cancer risk for high versus low levels of physical activity (relative risk (RR) 0.93, 95 % confidence interval (CI) 0.88-0.98). By comparison, case-control studies yielded a stronger, statistically significant risk reduction (RR 0.78, 95 % CI 0.66-0.94; p-difference by study design = 0.07). When focusing on cohort studies, physical activity summary risk estimates appeared to be more pronounced for consistent physical activity over time (RR 0.86, 95 % CI 0.76-0.97) than for recent past physical activity (RR 0.95, 95 % CI 0.90-1.01) or distant past physical activity (RR 0.95, 95 % CI 0.79-1.15, p-difference by timing in life of physical activity = 0.36). Physical activity summary risk estimates did not differ by smoking status or BMI. In conclusion, physical activity is not strongly associated with pancreatic cancer risk, and the relation is not modified by smoking status or BMI level. While overall findings were weak, we did find some suggestion of potential pancreatic cancer risk reduction with consistent physical activity over time.