AbstractTemporal alterations in the structure of hydropsychid guilds were studied along a regulated stream (Río Duratón, northern Spain) to examine the role of interspecific competitive interactions and environmental perturbations in determining species persistence. Discharge fluctuations of hypolimnial waters with a significant oxygen deficit were produced daily by Burgomillodo Dam. Under unregulated conditions (upstream sampling site), it is assumed that the coexistence of competing hydropsychid species is a function of species dominance (d') and resource limitation (R), as major causes of interaction strength among competitors, and the niche overlap (O) as a major cause of connectance among competitors, d' tends to increase with the number of subordinate species. Thus, and according to May's model of stability, a hydropsychid guild will tend to be unstable (by competitive displacement) if d'R(SO)0.5 > 1, where S is the number of competing species in the guild. The product d'R(SO)0.5 is used as a competition coefficient (β′). To facilitate the calculation of β′, the values of R and 0 are assumed to be 1 and 0.5, respectively. Riffle macrobenthic surveys were undertaken in June and December 1987, and in June 1988. Upstream from the dam, values of β′ were 0.599, 1080 and 0.656, and values of S were 7, 7 and 6. Downstream from the dam, β′ was less than unity during all sampling surveys, but S decreased at all downstream sampling sites (mainly just below the dam). Total density and total biomass were significantly higher downstream than upstream (except just below the dam). Hydropsyche pellucidula was dominant upstream, whereas H. siltalai became dominant downstream. Temporal variations in individual weights of H. bulbifera and H. exocellata downstream were different from those upstream. This fact is interpreted as a phenomenon of character displacement. It is concluded that interspecific competitive interactions upstream and environmental perturbations downstream determined species persistence in hydropsychid guilds.