Clarifying the thickness and its temporal variation rate of the mollic epipedon across cultivated lands in northeastern China is crucial for the protection of Mollisols and food security. However, accurately determining the thickness and its variation of the mollic epipedon is challenging due to the lack of comprehensive datasets and the spatial mismatch of sampling sites from different time periods. This study mainly aimed to determine the spatiotemporal variation in mollic epipedon thickness in the drylands of northeastern China over the past 40 years. We developed a method to identify the thickness of the mollic epipedon from 4410 soil profiles collected in the 1980s. In the 2020s, a rapid survey method was proposed to measure the thickness of the mollic epipedon. We then developed spatial predictive models of mollic epipedon thickness for the 2020s and 1980s, respectively, using random forest algorithms. Finally, we evaluated the thickness and variation of the mollic epipedon for drylands over the past 40 years by combining data from sampling sites and spatial distributions. Results indicate that both the identification method (R2 = 0.80, N = 69) and the rapid survey method (R2 = 0.91, N = 69) for mollic epipedon thickness perform well. The spatial prediction models also perform well for both the 2020s and 1980s, with R2 (RMSE) values of 0.94 (9.1 cm, N = 732) and 0.94 (8.8 cm, N = 4410), respectively. The spatial distributions indicate a decreasing trend in mollic epipedon thickness from north to south in both periods, with an average thickness of 45.0 cm in the 2020s and a thinning rate of 0.25 cm per year over the past 40 years. Findings from sampling sites support the spatial distribution of variation rate for mollic epipedon thickness (R2 = 0.78, N = 69). The thickness and variation rate of the mollic epipedon are related to soil type and slope position. The mollic epipedon thickness is largest for the soil type of Hapli-Udic Isohumosols and smallest for Hapli-Udic Argosols, with the thinning rate largest for the former soil type (0.38 cm/y). The thinning rate decreases with slope position, with average rates of 0.38 cm/y, 0.27 cm/y, and 0.13 cm/y for upper, middle, and lower slope, respectively. This is the first large-scale assessment of thickness and variation of the mollic epipedon in northeastern China over the past 40 years, utilizing extensive sampling data and machine learning methods. These findings provide valuable technical and data support for the conservation of Mollisols.