Large dams have been constructed on rivers to meet human demands for water, electricity, navigation, and recreation. As a consequence, flow and temperature regimes have been altered, strongly affecting river food webs and ecosystem processes. Experimental high‐flow dam releases, i.e., controlled floods, have been implemented on the Colorado River, USA, in an effort to reestablish pulsed flood events, redistribute sediments, improve conditions for native fishes, and increase understanding of how dam operations affect physical and biological processes. We quantified secondary production and organic matter flows in the food web below Glen Canyon dam for two years prior and one year after an experimental controlled flood in March 2008. Invertebrate biomass and secondary production declined significantly following the flood (total biomass, 55% decline; total production, 56% decline), with most of the decline driven by reductions in two nonnative invertebrate taxa, Potamopyrgus antipodarum and Gammarus lacustris. Diatoms dominated the trophic basis of invertebrate production before and after the controlled flood, and the largest organic matter flows were from diatoms to the three most productive invertebrate taxa (P. antipodarum, G. lacustris, and Tubificida). In contrast to invertebrates, production of rainbow trout (Oncorhynchus mykiss) increased substantially (194%) following the flood, despite the large decline in total secondary production of the invertebrate assemblage. This counterintuitive result is reconciled by a post‐flood increase in production and drift concentrations of select invertebrate prey (i.e., Chironomidae and Simuliidae) that supported a large proportion of trout production but had relatively low secondary production. In addition, interaction strengths, measured as species impact values, were strongest between rainbow trout and these two taxa before and after the flood, demonstrating that the dominant consumer–resource interactions were not necessarily congruent with the dominant organic matter flows. Our study illustrates the value of detailed food web analysis for elucidating pathways by which dam management may alter production and strengths of species interactions in river food webs. We suggest that controlled floods may increase production of nonnative rainbow trout, and this information can be used to help guide future dam management decisions.