Assessing food web structural properties and energy fluxes under changing hydrological regimes and water quality reveals the temporal patterns of ecosystem dynamics in shallow lakes. Here, we studied northern China’s largest shallow lake (Lake Baiyangdian) using quantitative food web models for five representative years (1958, 1980, 1993, 2009, and 2019). We analyzed the temporal patterns of food web structure and function by combining a Bayesian isotope mixing model with a food web energetics model. We further examined the temporal changes of unweighted and weighted food web topological attributes. Lake Baiyangdian changed from a detritus-based into a phytoplankton-based food web based on the relative contributions of basal food sources and energy flux distributions. The trophic position of fingerlings, large omnivorous fish, and carnivorous fish decreased with increasing eutrophication. The highest energy fluxes were from detritus to zooplankton and mollusks in 1958, from detritus and phytoplankton to zooplankton in 1980, 1993, and 2009, and from phytoplankton to zooplankton and detritus to mollusks in 2019. The highest total energy flux was in 1993, followed by 2019, with the lowest value in 1958. Unweighted food web metrics showed similar patterns. We observed more pronounced temporal variability in the node- and link-weighted food web metrics than in the unweighted metrics. In addition, hydrological factors (threshold, duration, reversals between high, and low water levels), eutrophication, and some water quality factors (chemical oxygen demand, dissolved oxygen, and pH) played important roles in the temporal changes of food web dynamics in Lake Baiyangdian. Our findings demonstrate the importance of integrating unweighted and weighted indicators to holistically comprehend how highly aggregated food webs respond to changing hydrological regimes and water quality, thereby improving management and restoration of shallow lake ecosystems.