Special mining methods and red soil lead to large-scale land degradation and desertification in ion-type rare earth (RE) mining areas. Therefore, it is crucial for ecological management and restoration of mining areas to accurately understand the evolution process of desertification. In this study, remote sensing Landsat images from 1986 to 2019 were used to extract desertified land information from the Lingbei mining areas, Dingnan County, Ganzhou, China. To improve the reliability of the experiment, samples selected from Google images were used for verification to compare the accuracy of the desertification difference index (DDI) model and random forest (RF) algorithm for extracting land desertification information. The results showed that compared with the DDI model, the overall accuracy and kappa coefficient of the RF model based on multiple features were improved by 7% and 9.37%, respectively, indicating its higher applicability. Spatiotemporal change analysis of desertification in the mining area showed that the total area of desertification in the mining area increased most rapidly during 1986–1994 and reached 60.75 km². The area of desertified land increased continuously from 1994 to 2004 and reached a maximum of 143.08 km² in 2004. The area of desertified land decreased by 50.27 km², but the severe desertified land (SDL) area increased by 1.69 km² during 2004–2011. The area of desertified land gradually declined and stabilized from 2011 to 2019. Analysis of the desertification process in mining areas under different disturbance conditions showed that the desertified land disturbed by RE mining was most severely damaged. There is still an area of 16.77 km² in the process of restoration, of which 2.24 km² belongs to the SDL level. Moderate desertified land (MDL) and light desertified land (LDL) have not been completely contained and require the attention of the relevant departments to ensure their timely reclamation.