An empirical model predicting probability of encountering levels of mercury in excess of 0.5 p.p.m. in lake trout (Salvelinus namaycush) is developed using logistic regression and two published data sets on food-web structures and elevated levels of mercury in sport fish. Probability of mercury contamination is a function of lake trout length and presence of pelagic forage fish. Presence of forage fish in a lake increased the odds that a lake trout of a given size would contain elevated levels of mercury in excess of eight times the expected level in lakes without pelagic forage fish. The model predicted a less than 5% probability of encountering elevated mercury levels in lake trout less than 30 cm in length from lakes with pelagic forage fish. In lakes where pelagic forage fish were absent, there was a less than 5% probability that lake trout under 45 cm would contain levels of mercury above 0.5 p.p.m.