Long-chain (n-3) PUFA (LC-PUFA) have been hypothesized to be beneficial in preventing pancreatic carcinogenesis, but the associations of fish or LC-PUFA intake with pancreatic cancer found in epidemiologic studies have been controversial and inconclusive. To estimate the overall association of LC-PUFA or fish intake with pancreatic cancer, we performed a systematic literature search of English-language articles using PubMed and EMBASE through February 2012 and reviewed the reference lists from retrieved articles. Prospective cohort or case-control studies that reported ratio estimates and corresponding 95% CI for the associations of fish or LC-PUFA intake and pancreatic cancer were selected. Independent data extraction was performed by 2 of the authors. The pooled associations were obtained by using a random-effects model. A database was derived from 9 independent cohorts that included 1,209,265 participants (3082 events) with a mean follow-up of 9 y and 10 independent case-control studies that included 2514 cases and 18,779 controls. Compared with those having the lowest fish consumption, the pooled RR of pancreatic cancer was 0.98 (95% CI: 0.86, 1.12) for those who had the highest fish intake from 8 cohort studies and was 0.96 (95% CI: 0.76, 1.21) from 9 case-control studies. We found similar results for LC-PUFA intake by combining data from 4 cohorts or 2 case-control studies. Our results do not support an overall inverse association of fish or LC-PUFA intake with risk of pancreatic cancer. Further studies that consider different species and preparation methods of fish, and additional adjustment for contaminants in fish, are warranted.