Persistence, a measure of presence or absence of species, and stability, an estimate of assemblage equilibrium measured as constancy in species ranks or densities, should both be considered when assessing temporal change in natural assemblages. We measured persistence and stability in two distinct animal assemblages frequently disturbed by natural and severe flooding events, the fishes and benthic invertebrates in a Sonoran Desert stream. A persistence index (derived from colonization/extinction analyses) indicates high persistence of fishes for several decades, while benthic invertebrates were persistent except in periods of severe flooding. Stability of taxon rankings (measured by Kendall's W) was high for both assemblages, even though absolute population sizes fluctuated. Fish populations resisted even the most severe flood disturbances, whereas benthic invertebrates were decimated by particularly frequent and intensive flooding. The latter were resilient, however, and quickly recovered due to life history characters favoring rapid postflood recolonization. Although absolute numbers of organisms varied through orders of magnitude, more general aspects of assemblage structure (species' presence or absence, and relative rankings) remained relatively constant despite repeated and potentially devastating natural perturbations. INTRODUCTION In their synthesis of ecological stability and persistence, Connell and Sousa (1983) emphasized the utility of a multifaceted approach in temporal analyses of community structure. Drawing upon the works of Holling (1973), Boesch (1974), Orians (1974) and Whittaker (1974), among others, they presented qualitative and quantitative criteria to be considered when assessing changes in community structure over time. Qualitatively, persistence of species focuses upon their continued presence, particularly with respect to potentially devastating forces. Quantitatively, stability is the relative constancy of species abundances over time despite disturbances. Stability may result from resistance, when relative species abundances are unchanged despite potentially disruptive forces, or resilience, rapid return to the former state following disturbance. Both persistence and stability should be considered when attempting to understand community pattern and process (Holling, 1973). The possibility of destructive disturbance is a necessary feature of communities that are used to investigate temporal change. . . for any situation to be included under the concept of stability . . .there must exist disturbing forces; without them, simple lack of change in numbers is of little interest (Connell and Sousa, 1983). Consequently, responses of natural communities to disturbance are central to current ecological research (Dayton, 1971; Sousa, 1979, 1984; Matthews, 1986), and disturbance provides a framework upon which mechanistic understandings of community structure and function may be developed. We examined persistence and stability of fish and benthic invertebrate assemblages in a Sonoran Desert stream that repeatedly experiences large, quantifiable natural dis-