Fish and omega-3 polyunsaturated fatty acids (PUFA) have been suggested to play a role in improving cancer prognosis. However, results from epidemiological studies remain inconsistent. Here we assess the association between dietary fish and/or omega-3 PUFAs intake and cancer prognosis with meta-analysis of observational studies. A systematic search of related publications was performed using PubMed and Web of Science databases. Hazard ratios (HR) and 95% confidence intervals (CI) were extracted and then pooled using a random-effect model. Potential linear and non-linear dose-response relationships were explored using generalized least squares estimation and restricted cubic splines. As a result, 21 cohort studies were included in our analysis. Compared to the lowest category, the highest category of fish intake was associated with a significant lower mortality in patients with ovarian cancer (n = 1, HR = 0.74, 95% CI: 0.57-0.95) and overall cancer (n = 12, HR = 0.87, 95% CI: 0.81-0.94). Marine omega-3 PUFAs intake rather than total omega-3 PUFAs intake showed significant protective effects on survival of overall cancer (n = 8, HR = 0.81, 95% CI: 0.71-0.94), in particular prostate cancer (n = 2, HR = 0.62, 95% CI: 0.46-0.82). Dose-response meta-analysis indicated a nonlinear and a linear relationship between fish intake, as well as marine omega-3 PUFAs intake, and overall cancer survival, respectively. In conclusion, our analysis demonstrated a protective effect of dietary fish and marine omega-3 PUFAs consumption on cancer survival.