Sediments of the St. Lawrence River at Cornwall, Ontario, Canada, are contaminated with mercury (Hg) due to industrial pollution. To better understand the risk to fish-consumers of Hg poisoning, patterns of Hg bioaccumulation in yellow perch were described at small spatial and temporal scales within a 10 km stretch of the river. Yellow perch (mathematically standardized to the average length of capture, 13.8 cm) from a contaminated upstream zone contained 2.3–3.6 times more mercury than fish from reference zones and a more contaminated zone only 2–3 km downstream. This was confirmed in two consecutive field seasons and with other fish species. Mercury concentrations in fish could not be accounted for by differences among study sites in fish length, age, condition factor, or growth rate. However, there were seasonal changes in the variance of the log(Hg)-size relationship and in fish size and total Hg concentrations. Possible reasons for the observed patterns could include differences among sites in: food web complexity, prey contamination, fish bioenergetics and competition, local sources of Hg, or fish movements into and out of the areas of contamination. This study demonstrated that Hg in sediments is not a reliable basis for estimating human or ecological risks. Fish tissue analysis provides a clearer and more direct measure of exposure of human consumers. Similarly, annual programs of regional-scale monitoring may overlook local-scale contamination and the change in risks of Hg exposure due to seasonal fish movements.