Increasing demand for river water now conflicts with an increasing desire to maintain riparian ecosystems. Efficiently managing river flows for riparian vegetation requires an understanding of the time scale of flow effects, but this information is limited by the absence of long-term studies of vegetation change in response to flow variation. To investigate the influence of short- and long-term flow variability and dam operation on riparian vegetation, we determined the occurrence of 107 plant species in 133 permanent plots of known inundating discharge along the Gunnison River in Colorado on five different occasions between 1990 and 2013. Individual species moved up and down the gradient of inundating discharge coincident with increases and decreases in mean annual flow, and the correlations between flow and species occurrence were strongest when flows were weighted by time before vegetation sampling with a median half-life of 1.5 years. Some tall, rhizomatous, perennial species, however, responded to flows on a longer time scale. Logistic regression of species occurrence showed a significant relation with inundation duration for 70 out of 107 species. Plot species richness and total vegetative cover decreased in association with desiccation at low inundation durations and with fluvial disturbance at high inundation durations. Within-plot similarity in species occurrence between years decreased strongly with increasing inundation duration. Moderate inundation durations were dominated by tall, rhizomatous, perennial herbs, including invasive Phalaris arundinacea (reed canary grass). Over the 23-year study period, species richness declined, and the proportion of rhizomatous perennials increased, consistent with the hypothesis that decreases in flow peaks and increases in low flows caused by flow regulation have decreased establishment opportunities for disturbance-dependent species. In summary, annual-scale changes in vegetation were strongly influenced by flow variation, and decadal-scale changes were influenced by decreases in fluvial disturbance from upstream flow regulation beginning decades prior to the onset of this study.