As wildfire regimes shift, resource managers are concerned about potential threats to aquatic ecosystems and the species they support, especially fishes. However, predicting fish responses can be challenging because wildfires affect aquatic ecosystems via multiple pathways. Application of whole-ecosystem approaches, such as food web modeling, can act as heuristic tools that offer valuable insights that account for these different mechanisms. We applied a dynamic food web simulation model that mechanistically linked stream trophic dynamics to the myriad effects that wildfires can have on aquatic and riparian ecosystems at a local stream reach-scale. We simulated how wildfires of different severity may influence short- (months to years) and long-term (years to decades) periphyton, aquatic invertebrate, and fish biomass dynamics in forested headwater streams of the western Pacific Northwest (USA). In many cases, wildfire increased modeled periphyton, invertebrate, and fish biomass over both short- and long-time periods. However, modeled responses varied extensively in their direction (that is, positive or negative), magnitude, and duration depending on fire severity, time since fire, and trophic level. The shapes of these response trajectories were especially sensitive to predicted wildfire effects on water temperature, canopy cover, riparian shading, and instream turbidity. Model simulations suggest a single fire could result in a wide range of aquatic ecosystem responses, especially in watersheds with mixed burn severity. Our analysis highlights the utility of whole-ecosystem approaches, like food web modeling, as heuristic tools for improving our understanding of the mechanisms linking fire, food webs, and fish and for identifying contexts where fires could have deleterious impacts on fishes.