Four 0.5-m plankton nets (one each of 0.355-, 0.450-, 0.560-, and 0.750-mm mesh) were used to collect larvae of the alewife (Alosa pseudoharengus) and rainbow smelt (Osmerus mordax) in northwestern Lake Huron and the St. Marys River. The number of larvae collected varied inversely with mesh size and the average length of larvae varied directly with mesh size. Numbers of each species caught in 0.355- and 0.450-mm mesh nets were significantly greater (P < 0.01) than the numbers caught in 0.560- and 0.750-mm mesh nets. Numbers of alewives caught were significantly greater (P < 0.01) in the 0.355- than in the 0.450-mm mesh, but numbers of rainbow smelt caught in the two meshes were about equal. Between the two larger-mesh nets, numbers of each species caught did not differ significantly. Nets with mesh larger than 0.355 mm were less effective at catching recently hatched larvae, and this reduced effectiveness accounted wholly for the smaller numbers caught in the 0.450-mm mesh net. Smaller numbers were caught in the 0.560- and 0.750-mm mesh nets not only because they caught fewer recently hatched larvae but also because they failed to retain older, larger larvae. Information on the effect of mesh size and towing speed on catches of larvae is important for making accurate estimates of larval densities.