ABSTRACT: A divergence in sieve size protocols for washing samples has arisen among shallow- and deep-sea benthic biologists, which now affects comparability across the 2 environments. This has come about as a result of a perception of smaller body sizeamong deep-sea benthic organisms. Two box-core samples from ~1900 m depth were examined to see how different sieve size affects estimation of biomass, abundance and diversity of macrofauna in the deep sea. Expressed as cumulative retentions, the coarsestsieve (1 mm mesh) retained 94% of the biomass retained in the finest sieve (0.25 mm), confirming the expectation that rarer, large-sized individuals contribute most to biomass of macrobenthos. There were small increases when progressively finer sieveswere used, but the rate of increase declined markedly from 0.425 to 0.25 mm mesh. Numbers of individual organisms increased through the series, with a marked increase from 0.5 to 0.425 mm mesh; numbers of species did not increase as rapidly, but showed asimilarly high rate between 0.5 and 0.425 mm mesh. The proportions of the total biomass in coarser sieves, and the total number of specimens retained by the finest sieve were rather similar to those shown in previous studies of inshore and continentalshelf benthos. However, the retention of deep-sea species by the 0.5 mm sieve (86%) was about the same as the percentage of inshore species retained by the 1.0 mm sieve (at this sieve size only 54% of deep-sea species were retained). Although the finersieves in both shallow- and deep-sea samples include many postlarval stages, this differing response for species must reflect smaller size of some adult macrobenthos in the deep sea. The effect on diversity depended on index used, with Shannon¹s indexincreasing down to the 0.5 mm mesh, but showing a variable response in the finer sieves. This probably mostly reflects increased retention of small-sized juveniles of species already retained, causing increasing unevenness in species abundances withsmaller sieve size, rather than addition of small-sized species not previously retained in coarser sieves. Expected numbers of species calculated by rarefaction seems much less affected, with the degree of overlap of rarefaction curves reflecting relativesimilarity in distribution of species abundances. Degree of overestimation (of expected number of species against number of individuals along the curve) is greatest for the 1 mm mesh, even if the end-points of the curve show decreasing disparity, from 1to 0.25 mm mesh, in total retention of species as well as individuals. It is argued that, as would apply in shallow water, the great dependence of evenness on sieve size makes shifts in distribution of species abundances caused by disturbance moredifficult to detect against the baseline, unless a sufficiently small sieve has been used. When sieve retention was separately considered for the major taxonomic groups of macrobenthos (Polychaeta, Peracarida, Mollusca), differing patterns in relation tosieve size were found in terms of abundance and species. No single group can therefore serve as a reliable proxy for the total macrobenthic assemblage.