Marine biodiversity is being rapidly redistributed in response to ocean change, yet globally consistent datasets that resolve suitable habitat across both depth and time remain scarce. Here, we present a multi-decadal, depth-resolved Area-of-Habitat (AOH) dataset for a globally distributed subset of marine species assessed by the IUCN Red List. We integrated approximately 45.5 million occurrence records with IUCN expert range maps and physical and biogeochemical ocean reanalysis data spanning 1993–2025. Species-specific MaxEnt models were successfully fitted for 5,619 species primarily fishes (4,815 species), together with reef-forming corals, molluscs, lobsters, marine mammals, marine reptiles, mangroves and seagrasses. The fitted models were projected across 50 standard depth layers for 1995, 2000, 2005, 2010, 2015, 2020 and 2025. Phytoplankton, zooplankton and macroalgae were not included. Model predictions were constrained to the original IUCN ranges to generate continuous, range-constrained estimates of environmental suitability, operationally defined here as marine AOH. The resulting dataset comprises species-level layers and aggregated products resolved by depth and year. Internal consistency assessments showed that predicted suitability was generally higher at species occurrence locations across the evaluated species and depth layers. Independent comparisons with global coral-reef distributions and FishGlob scientific survey observations further supported the ecological plausibility and spatial realism of the resulting AOH estimates.</p> <p>Aggregated analyses revealed pronounced vertical differentiation in marine AOH, with shallow and deeper waters exhibiting distinct spatial concentrations and temporal trajectories. By integrating geographic space, ocean depth and multi-decadal environmental change within a consistent modelling framework, this dataset provides a standardized resource for investigating marine habitat redistribution, identifying persistent areas of high suitability and potential climate refugia, and evaluating the correspondence between suitable habitats and marine protected areas in a changing ocean. The dataset is publicly available at <a href="https://doi.org/10.5281/zenodo.21933262">https://doi.org/10.5281/zenodo.21933262</a> (Yu and Li, 2026).