Climate change models predict increasing frequency of flood and drought in Great Plains streams.  The ultimate influence of these predicted changes on stream ecosystem structure and function may depend on community structure, which is variable in prairie streams.  We used experimental streams to test the interactive effects of flood frequency (12-day, 24-day, or no-flood) and the presence or absence of an omnivorous fish (Cyprinella lutrensis [8.9/m²]) on invertebrate abundance and biomass.  Each stream consisted of a 2.54/m² pool connected to a 0.84/m² riffle.  Floods were simulated with a pressure hose.  Separate field experiments validated that experimental streams were good analogs for natural stream reaches. Invertebrates were sampled on days 1, 6, 30, 42, and 54 using a 0.018/m² corer.  Invertebrate abundance and biomass were consistently greater in no-flood treatments and lowest in 12-day flood interval treatments.  Invertebrate abundance in no-flood and 24-day flood treatments was greater in fish treatments compared to no-fish treatments.  Greater abundance was primarily due to significant increases in small crustaceans and was correlated with increased primary production in fish treatments.  There was no fish effect in the 12-day flood treatments, supporting hypotheses that frequent disturbance negates the influence of biotic interactions.