a b s t r a c tChina has experienced socio-economic transitions in recent decades, featuring a large amount of rural labor migrating into urban areas. As a response, rural land use has been restructured across large areas. In this process, vast amounts of land have been abandoned due to labor loss, especially in mountainous areas. Understanding how land attributes determine which land parcels are abandoned is essential to sustainable development of rural areas. Here we examine how two main farmland parcel attributes, farmland-to-housing distance and land quality, by affecting farming costs and benefits, and thereby profits, determine farmland abandonment. We constructed a semi-empirical crop profit model based on output benefits and input costs derived from household survey data from mountainous Wulong County, Chongqing Province, China. With this model, crop profits can be estimated and in turn the farmland-tohousing distances at which farming profits diminish to zero ("zero profit distance", abbreviated as "ZPD") can be identified. Based on the hypothesis that land will be abandoned when cultivation cannot make profits, ZPD values can be used to predict patterns of abandonment. At current price levels, the ZPD values are 5.94 km, 3.84 km, 2.52 km and 1.48 km for first-class, second-class, third-class and fourthclass land, respectively. Overall, the observed occurrence of land abandonment in relation to farmland-to-housing distance and land quality is concordant with the model's predictions, supporting the underlying hypothesis that land will be abandoned when the farmland-to-housing distance exceeds the ZPD. These findings may serve as an important tool for predicting land abandonment and identifying countermeasures for mountainous areas.