<p>Assessment of land use and land cover, change detection mapping, and further integration with land capability classes are effective tools for sustainable watershed planning. In this context, the present study was conducted in the Fatehpur watershed of Gaya district, Bihar, to investigate the temporal and spatial dynamics of LULC over the past two decades (2004–2024) and its relation to land capability classes. Supervised LULC classification was performed with a focus on maximum likelihood classification to ensure high accuracy in mapping and validated through field surveys, which revealed significant LULC changes between 2004 and 2024. Forest cover has decreased by 11.3% and agricultural land by 9.6% in the past 20 years, with corresponding increases in fallow and non-agricultural land. The Kappa coefficient values of the LULC classification were greater than 0.7, which indicated that the categorization implies significantly better agreement than chance alone. In comparison to the present scenario, the data on the change matrix showed that open scrub (7.9%) and forest (7.1%) lands were converted to crop land. Furthermore, cross-analysis of land capability classes with LULC changes showed that the majority of cultivable land was classified as moderately suitable land. Overall results showed that deforestation has reached 11% of the study area over the past two decades. Therefore, agroforestry, planting trees on farm boundaries, or growing trees on lands that could not be used for agricultural purposes, is an important land use option that can ameliorate soil erosion and restore degraded forests or wastelands in the study area.</p>