GCTB provides a global dataset of urban and town boundaries from 1985 to 2022 at a 30 m spatial resolution. It delineates the spatial extents of cities and towns over a 27-year period, integrating global coverage with population-informed functional classification. Data are provided at five-year intervals before 2000 (i.e., 1985, 1990, and 1995) and annually from 2000 to 2022. Each temporal layer includes two vector files: Cities_Year (city boundaries) and Towns_Year (town boundaries), where Year refers to the year of observation (e.g., Cities_1985 ,Towns_1985). All data layers are projected in the WGS-84 geographic coordinate system (EPSG:4326) and delivered in vector polygon format. Specifically, urban boundaries were delineated by combining kernel density estimation (KDE) and cellular automata (CA) algorithms to ensure spatial continuity. Cities and towns were functionally distinguished based on population size, enabling a consistent classification of settlements across regions. Compared to high-resolution manually interpreted datasets, such as the Atlas of Urban Expansion (AUE), GCTB achieved high consistency (R² = 0.889 in 1999 and 0.941 in 2014) and the lowest RMSE among all available global urban boundary datasets.