Much of the discussion of high deep-sea diversity has assumed that asymmetric inter-specific competition will rapidly lead to the elimination of many species unless restrained by extensive differentiation of niches, or the action of predation and/or environmental disturbance. This is true for some habitats, including rocky shores. However, experimental studies indicate that marine soft sediment communities do not function like this. In shallow-water sediments, competition is usually symmetric and relatively weak. Asymmetric competition that leads to the elimination of one species by another on time scales shorter than one generation is rare, apart from interactions that involve large-scale modification or disturbance of the sediment. Competition is therefore relatively unimportant as a process structuring communities and the impact of predation is usually to reduce rather than enhance diversity. These results have been largely ignored by the literature on deep-sea diversity. If deep-sea communities function in similar ways, coexistence of many species within small areas on short time scales does not need further explanation. We do, however, need to explain why abundances of all species remain bounded on large spatial scales and time scales of several generations. The model of diversity maintenance proposed by Huston (1979) and applied to the deep sea by Rex (1983) achieves this by implicit intra-specific, density-dependent processes that increase the population growth rate of species that have become locally rare. This would give robust maintenance of diversity, but there is no evidence for density dependence of this type in the deep sea, and no plausible mechanisms by which it could occur. Alternative models require either spatial heterogeneity on a scale much larger than that envisaged by the grain-matching hypothesis or the placing of a cap on the abundance of common species, perhaps by frequency-dependent predation. Arbitrating between these possibilities will require assessments of the population dynamics and spatial distribution of individual species on spatial and temporal scales much greater than those usually considered in the deep sea.