Ecosystem regime shifts can alter ecosystem services, affect human well-being, and trigger policy conflicts due to economic losses and reductions in societal and environmental benefits. Intensive anthropogenic activities make the Sea of Marmara ecosystem suffer from nearly all existing available types of ecosystem pressures such as biological degradation, exposure to hydrological processes, nutrient and organic matter enrichment, plastic pollution, ocean warming, resulting in deterioration of habitats. In this study, using an integrated ecosystem assessment, we investigated for the first time the historical development and ecosystem state of the Sea of Marmara. Multivariate analyses were applied to the most comprehensive and unique long-term data sets of 9 biotic and 15 abiotic variables for ecosystem state and drivers respectively, from 1986 to 2020. Observed changes were confirmed by detecting shifts in the datasets. The Sea of Marmara ecosystem was classified into three regimes: i) an early initial state regime under the top-down control of predatory medium pelagic fish and fisheries exploitation until mid-1990s, ii) a transitional regime between mid-1990s and mid-2010s as from ecosystem restructuring, and iii) an alternate state late regime with prevailing impacts of climate change from mid-2010s until 2020. During the 20 years transitional regime, three different phases were also characterized; i) the 1st phase between mid-1990s and early 2000s with its gradual change in ecosystem state from a decrease in predators and significant shift in physical drivers of the ecosystem, ii) the 2nd phase between 2000 and mid-2000s with a strong shift in ecosystem state, an ongoing increase in climate indices and fishing mortality, and a gradual decrease in water quality; and iii) the 3rd phase between mid-2000s and mid-2010s with the reorganization of the ecosystem dominated by small pelagic fish and ameliorated water quality. During late regime, we observed that most of the biotic variables, mainly fish biomass, and climate variables did not return to their initial state despite the improvement in some abiotic variables such as water quality. We identify these observed changes in the SoM ecosystem as a non-linear regime shift. Finally, we also developed concrete suggestions for improved regional management.