Fish provide healthy protein as well as recreational and cultural benefits, but can also contain mercury (Hg), polychlorinated biphenyls (PCBs), and other contaminants that have adverse effects on humans and other organisms, particularly developing fetuses. Recently, some authors have suggested that a molar excess of selenium (Se) [e.g., selenium/mercury (Se/Hg) molar ratio >1] confers protection from Hg toxicity derived from fish consumption. Herein, we review our studies of Hg and Se in freshwater, marine, and commercial fish (mainly marine), examining the following: (1) whether and how Se/Hg molar ratios vary among species; (2) whether and how the molar ratios vary within species; (3) whether the molar ratios differ between freshwater and saltwater fish; (4) whether mean molar ratio values provide a reliable indication of potential risk to fish consumers; and (5) whether mean Se/Hg molar ratios are sufficiently constant (e.g., low variation) to allow for use in risk assessment, risk management, or risk communication. In saltwater fish, mean Se/Hg molar ratios varied from 0.3 in mako shark to 68.1 in whiting. For freshwater fish, the mean ratios varied from 0.68 in bowfin to 20.8 in black crappie. Commercial seafood (mainly saltwater) showed great variation in ratios; shrimp and scallops had very high ratios. There was somewhat less variability in the ratios for freshwater fish compared with the fish from saltwater, but there was no overall predictable difference in variation in Se/Hg molar ratios. For both saltwater and freshwater fish, some species with mean molar ratios above 1 had a significant proportion of individual fish with molar ratios below 1. Overall, this indicates great variation in measures of central tendencies and in measures of dispersion. We suggest that relying on the Se/Hg molar ratio as a method of predicting reduced risk from Hg toxicity is problematic because of the great variation among and within fish species, and the variation is not predictable because Hg varies by season, size of the fish, and location of the fish (which is not available for commercial fish). With the high variation in ratios, and low predictability, the ratios are currently not useful for risk assessment and risk management, and vulnerable individuals cannot rely on mean Se/Hg molar ratios for protection from Hg toxicity.