This paper compares collection characteristics of #2-(363 pm), #lo-(156 km), and #20-(76 pm) mesh conical plankton nets: dimensions were 50-cm diameter by 1.6-m long. The #2-mesh net severely underestimated the abundances of Lake Michigan copepods and cladocerans with the exception of the largest species (Limnocalanus macrurus).Zooplankton abundance estimates were more similar for the #lo-and #20-mesh nets collections. Nauplii, however, were severely undersampled by the #IO-mesh net with abundance estimates approximately 8 to 12 times lower than for the #20-mesh net collections. Most other larger zooplankton were 50% more abundant in the 20-mesh net collections than in the #IO-mesh net collections: such consistent differences occurred despite large variations in taxa size. This indicates that a sampling bias occurred other than the loss of zooplankton through the meshes of the #lO net. We hypothesize that, by incorrectly locating the flowmeter in the mouth of the plankton net, we underestimated the volume of water filtered by the easily-clogged #20-mesh net and therefore overestimated taxa abundances. We conclude that the #IO-mesh net provided accurate estimates of microcrustacean zooplankton abundances except for nauplii. The #IO-mesh net used in our study had a filtration area ratio of 3.06 and operated at a calculated average filtration efficiency of 98%. The #20-mesh net had a filtration area ratio of 1.86 and operated at calculated average filtration efficiencies ranging from 64.7% (41.7 m station) to 79.6% (6.3 m station). Calculations are presented which show how the filtration efficiencies of the nets used in our study could be improved by net redesign