Harmful algal blooms (HABs) are a recurring problem in many temperate large lake and coastal marine ecosystems, caused mainly by anthropogenic eutrophication. Implementation of agricultural conservation practices (ACPs) offers a means to reduce non-point source nutrient runoff and mitigate HABs. However, the effectiveness of ACPs in a changing climate remains uncertain. We used an integrated biophysical modeling approach to predict how Lake Erie cyanobacterial HAB severity (bloom biomass) may change under several climate and ACP implementation scenarios, using western Lake Erie and its largely agricultural watershed as our study system. An ensemble of general circulation model projections was used to drive spatially explicit land use and hydrology models of the Maumee River watershed, the output of which informed a predictive model of Lake Erie HAB severity. Results show that, in the absence of changes in ACPs, the frequency of severe HABs is projected to increase during coming decades, owing to increased inputs of nutrients from the watershed. These anticipated increases are due to increased total precipitation and more frequent higher-magnitude rainfall events. While further implementation of ACPs appears capable of reducing severe HAB events, widespread implementation would be necessary to reduce HAB severity below current management targets. This study highlights how continued climate change will only exacerbate the need for land management practices that can reduce nutrient runoff in agriculturally dominated ecosystems, such as Lake Erie. It also shows how interdisciplinary, biophysical modeling approaches can help identify strategies to mitigate HABs in the face of anthropogenic stressors.