PURPOSE: To investigate the association between fish consumption and mortality in 36,003 Chinese. METHODS: A case-control study collected 81% of all deaths of those aged 30+ from all four Hong Kong death registries in 1998. Relatives registering the deaths provided demographic, dietary and other lifestyle data for the deceased (case) and a similarly aged living person (control). Causes of death were provided by the Department of Health. Logistic regression was used to calculate the mortality odds ratios (ORs) for fish consumption adjusting for potential confounders in the 23,608 cases and 12,395 controls. RESULTS: Compared with the lowest fish consumption of less than or equal to three times a month, higher consumption of one to three times a week was associated with lower mortality ORs (95% confidence interval [CI]) of 0.75 (0.62-0.89) for all-cause, 0.66 (0.48-0.92) for ischemic heart disease (IHD), 0.70 (0.50-0.98) for stroke, 0.66 (0.53-0.82) for cancer, but not for injury and poisoning. The highest level of fish consumption of greater than or equal to four times a week also reduced mortality with ORs (95% CI) of 0.80 (0.68-0.94) for all-cause and 0.63 (0.47-0.85) for IHD. CONCLUSIONS: Fish consumption significantly reduced mortality from several causes in this sample. Further longitudinal studies to confirm the association are needed.