Selectivity curves for three species of fish taken in South Bay, Manitoulin Island, and in Georgian Bay during 1954–59 with gangs of nylon gillnets differed with respect to mode, skew, and kurtosis. A graphic method was developed to describe one aspect of body shape, namely the relation between girth and distance from the snout, which would affect selectivity. Surprisingly, the shapes of mature (but not gravid) and immature perch and suckers of both sexes could be described by the same curve. Hence, differences in body shape did not account for differences in the selectivity of these species. In contrast, the taper for gravid female perch was markedly different from that for nongravid, as might be expected. Catches in which the positions where caught were carefully identified showed that most of the fish were caught either on the opercle or near the position of maximum girth. That the fish could be caught at different positions accounted for most of the skew in the selectivity curves based on maximum girth. Selectivity curves plotted on the basis of girth where caught, rather than on maximum girth or length, show that efficiency of capture was maximal when the girth of the fish was 1.0–1.2 times as great as the perimeter of the mesh. Efficiency was negligible at girth/perimeter ratios smaller than 0.8 or 0.9 and it declined rapidly at ratios greater than 1.2. However, a few fish were taken at ratios up to 1.5 or 1.6.