Long-term studies were used to examine (1) contrasting time scales and mechanisms of structural variations within two benthic communities and (2) the usefulness of long data sets for evaluating human impact. A 10-year study of a San Francisco Bay mudflat, the details of which are reported elsewhere, has revealed large short-term (on the order of months) variations in species abundances within a community composed predominantly of opportunistic species. The study site, located in a highly urbanized estuary, is subject to the influence of a nearby sewage-treatment facility. However, rapid changes in population size of the common species, in part due to periodic natural habitat disturbance, impedes the detection of anthropogenic influences on community composition. Only a very long-term data set may provide evidence of progressive change. Data collected for a 20-year period on the benthic community at 200 m depth in the main basin of Puget Sound, an environment subject to little apparent habitat disturbance show that numerical abundance of the common species can also change markedly. Here, however, numerical dominance shifts from one species to another at irregular, multiyear intervals. Recent increases in two heretofore rare species, and a significant increase in total numbers of individuals suggest that long-term changes may be occurring in this community. These two long-term data sets demonstrate the importance of measuring both the amplitude and the periodicity of fluctuations in population size of aquatic species as well as long-term fluctuations and patterns in environmental factors before attempting to demonstrate the effect of anthropogenic influences on aquatic communities. The results of these studies also demonstrate the usefulness of long-term data sets for revealing the potential importance of interactions among species in determining abundance patterns in the soft-bottom benthos.