Abstract: Coastal wetlands are fragile ecological zones formed by land-sea interactions. A thorough understanding of the impact of hydrological changes on their area change is therefore essential for effective ecological protection. Using the Qilihai coastal wetland in Tianjin as a case study, this research employed the Google Earth Engine (GEE) platform and an improved KT₃ algorithm to interpret Landsat imagery, thereby mapping changes in wetland extent from 1985 to 2020. By integrating ground-based hydrological and meteorological data, we quantified the wetland water balance and analyzed the driving effects of water cycle components on its evolution. The results showed that the wetland area exhibited a ‘W’-shaped fluctuation during 1985—2020, with a minimum of 14.95 km² in 2001 and a maximum of 69.92 km² in 2013. Variations in water cycle components were the dominant driver of wetland area change. In wet and normal years, net anthropogenic water inputs of 0.34×10⁸ to 1.26×10⁸ m³ were sufficient to maintain a positive water balance in the wetland. In dry years, however, anthropogenic net input declined to −0.52×10⁸ to −0.46×10⁸ m³, causing the wetland to shift to a negative water balance. Agricultural water diversion was the main outflow component, accounting for 28.7% to 32.3% of total outflow, with the highest proportion of 32.3% occurring in 1989, which reduced water availability for ecological use and consequently decreased wetland recharge. The Qilihai wetland is highly dependent on external water supply, and its shrinkage is characterized as water-deficit-driven degradation. Its interaction with groundwater has shifted from functioning as a seasonal discharge zone to becoming a recharge source. These findings provide a scientific basis for understanding the hydrological degradation mechanisms of coastal wetlands and for improving ecological water allocation and wetland restoration strategies.