Despite increasing efforts and investment in mangrove conservation, mangrove cover continues to decline globally. The extent to which protected area (PA) management effectively prevents mangrove loss globally across differing management objectives and governance types is not well understood. We combined remote sensing data with PA information to identify the extent and the drivers of mangrove loss across PAs with distinct governance types and protection levels based on categories developed by the International Union for Conservation of Nature (IUCN). Mangrove loss due to storms and erosion was prevalent across all governance types and most IUCN categories. However, the extent of human-driven loss differed across governance types and IUCN categories. Loss was highest in national government PAs. Private, local, shared arrangement, and subnational government agencies had low human-driven mangrove loss. Human-driven loss was highest in PAs with the highest level of restrictions on human activities (IUCN category I) due to mangrove conversion to areas for commodity production (e.g., aquaculture), whereas PAs that allowed sustainable resource use (e.g., category VI) experienced low levels of human-driven mangrove loss. Because category I PAs with high human-driven loss were primarily governed by national government agencies, conservation outcomes in highly PAs might depend not only on the level of restrictions, but also on the governance type. Mangrove loss across different governance types and IUCN categories varied regionally. Specific governance types and IUCN categories thus seemed more effective in preventing mangrove loss in certain regions. Overall, we found that natural drivers contributed to global mangrove loss across all PAs, whereas human-driven mangrove loss was lowest in PAs with subnational- to local-level governance and PAs with few restrictions on human activities.