Background <p>Wildfire can influence erosion regimes and alter sediment delivery to river systems, which can change the sediment size composition of rivers. Understanding how wildfire impacts streambed grain size distributions is essential in understanding broader wildfire impacts on aquatic habitat and downstream water resources. In Western Washington, Oregon, and Northern California, the Aquatic and Riparian Effectiveness Monitoring Program (AREMP) has been collecting ongoing monitoring data since 2002. Since the beginning of the monitoring program, approximately 37% of AREMP monitored sites have experienced one or more wildfire within their contributing watersheds. In this study, we used machine learning models to analyze AREMP sites within the Cascades and Klamath regions to assess (1) how the presence and severity of wildfire explain streambed sediment size in western Pacific Northwest (PNW) watersheds and (2) how watershed characteristics influence the effects of wildfire on streambed sediment across two distinct regions in the PNW.</p> Results <p>We found distinct patterns in how fire may be influencing instream sediment size in the Cascades compared to the Klamath, which could in part be explained by differences in fire regime and land management legacies between the two regions. In the Cascades, post-fire canopy cover and instream large wood seem to be driving differences in observed post-fire stream sediment compared to expected unburned conditions, while soil and lithology characteristics are the main drivers of difference in the Klamath region. Overall, we found that how streams respond to wildfire in the Cascades might be controlled by factors that influence sediment transport dynamics, while watershed response to fire in the Klamath may be controlled by hillslope sediment production.</p> Conclusions <p>Our findings offer insight into which landscape controls influence whether aquatic habitat might be vulnerable to or resilient to change post-fire, across different regions in the PNW.</p>