We compare estimates of biomass and communlty structure based upon depth-stratified sampling from 0 to 900 m at a site on the continental slope south of Tasmania. Australia, using an EK500 acoustic system with vessel-mounted and deeply towed 38 kHz split-beam transducers and a midwater trawl with opening-closing codend system. Multivariate analyses of the target strength (TS) distributions and of dominant species in the tows provided consistent views of community structure: an epipelagic communlty that was differentiated and mid-depth and deep communities not differentiated on a die1 basis. Peaks in the TS distributions and the estimated TS of key biological groups were congruent, although the in situ TS data indicated that fishes with swimbladders in the size range of myctophids have lower TS (up to 5 dB) than previously predicted for physoclists, possibly due to the smaller volume of gas-filled swimbladders among mesopelagic fishes. There was a progressive increase in modal TS distributions with depth, consistent with the larger mean s~z e of both squids and fishes in the deeper strata. Bionlass was lowest in the dayt~me near-surface stratum, but there were no other trends of biomass with depth. The acoustic estimate of biomass based upon the trawl-derived assessment of community composition was 7-fold higher than the trawl-based volume-swept biomass estimate. This acoustic biomass estimate was 3-fold less than one based upon an assessment of community composition derived from TS distributions. The higher acoustic biomass estimates are consistent with regional estimates of primary production and simple trophodynamic calculations.