Wetlands are crucial to diverse societies and ecosystems but are threatened globally. Natural wetlands in the middle and lower reaches of the Yangtze River Basin (MLRYRB) have long been exploited to meet the increasing food and energy demands, especially in recent decades. However, how their landscapes and hydrological patterns changed over time was poorly qualified, which was crucial to conserving biodiversity and ecosystems. In this study, we proposed a new wetland classification procedure taking advantage of the shape, topography, landscape patterns, and phenology to map the distribution of six wetland types in the MLRYRB in 1990 and 2020 with Landsat images. The transitions among different wetland types and associated driving factors were examined. Artificial wetlands expanded by 3907.6 km2 while natural wetlands decreased by 423.2 km2, resulting in a net area increase of 3484.4 km2. Farmland reclamations, aquacultural activities, urban expansion, and dam constructions induced 70 % of the area's loss of natural wetlands. Meanwhile, spatiotemporal patterns in wetland inundation dynamics were determined with agglomerative hierarchical cluster analysis. Widespread transitions from perennial water to seasonal wetlands were detected. Notably, 47.8 % of the natural wetlands experienced significant drying trends, and as a result, the mean yearly inundated periods for natural wetlands decreased from 303 days in 1990 to 272 days in 2020. Given the substantial structural alterations, wetland destruction was more severe than the area reflected. Our results suggested that category-based changes in wetland areas, inundation patterns, associated driving forces, and underlying ecological consequences should be comprehensively assessed to conduct scientific conservation and prevent “paper offsets.”