Arable land reclamation in arid zones has led to many conflicts between the use of artificial oases and natural ecosystems. An understanding of the processes used to determine arable land dynamics and their drivers can assist policy makers with oasis management. Therefore, we propose a GIS-based spatiotemporal trajectory analysis model to explore the temporal dynamics of arable land in the Tarim River Basin between 1985 and 2022 and to reveal the driving factors behind these changes. We first classified five trends of arable land changes: stable arable land, stable nonarable land, reclaimed land, abandoned land, and ephemeral arable land. We subsequently identified topographical factors, soil quality, nonirrigability, and inaccessibility as the driving factors of arable land changes, employing redundancy analysis to determine the relative significance of these factors. The results revealed that a total of 1.9 x 104 km2 of land was reclaimed during the study period, whereas 3.9 x 103 km2 of arable land was abandoned. Soil quality is the most important positive factor affecting stable arable land. Compared with stable arable land, nonirrigability has a reduced effect on reclaimed arable land, likely due to advances in irrigation technology that have lessened the dependence on irrigation conditions. Nonirrigability and inaccessibility have the greatest negative impacts on abandoned arable land, indicating that much of the abandonment in the Tarim River Basin is due to urban sprawl. This study is crucial for accurately assessing the drivers of arable land trajectories and for providing guidance for agricultural development, land management, and water conservation in arid regions.