The Lake Michigan fish community has been monitored since the 1960s with bottom trawls, and since the late 1980s with acoustics and midwater trawls. These sampling tools are limited to different habitats: bottom trawls sample fish near bottom in areas with smooth substrates, and acoustic methods sample fish throughout the water column above all substrate types. We compared estimates of fish densities and species richness from daytime bottom trawling with those estimated from night-time acoustic and midwater trawling at a range of depths in northeastern Lake Michigan in summer 1995. We examined estimates of total fish density as well as densities of alewife Alosa pseudoharengus (Wilson), bloater Coregonus hoyi (Gill), and rainbow smelt Osmerus mordax (Mitchell) because these three species are the dominant forage of large piscivores in Lake Michigan. In shallow water (18 m), we detected more species but fewer fish (in fish/ha and kg/ha) with bottom trawls than with acoustic-midwater trawling. Large aggregations of rainbow smelt were detected by acoustic-midwater trawling at 18 m and contributed to the differences in total fish density estimates between gears at this depth. Numerical and biomass densitites of bloaters from all depths were significantly higher when based on bottom trawl samples than on acoustic-midwater trawling, and this probably contributed to the observed significant difference between methods for total fish densities (kg/ha) at 55 m. Significantly fewer alewives per ha were estimated from bottom trawling than from acoustics-midwater trawling at 55 m, and in deeper waters, no alewives were taken by bottom trawling. The differences detected between gears resulted from alewife, bloater, and rainbow smelt vertical distributions, which varied with lake depth and time of day. Because Lake Michigan fishes are both demersal and pelagic, a single sampling method cannot be used to completely describe characteristics of the fish community.