Soil detachment by concentrated flow is a crucial component of soil erosion, and variations in rainfall-induced soil physical crust traits significantly influence soil detachment. In this study, to investigate the variation in soil detachment with rainfall-induced soil physical crust, rainfall simulations with a rainfall intensity of 1.5 mm m in-1and rainfall durations of 0 (CK), 5 (R5), 10 (R10), 15 (R15), 20 (R20), and 30 min (R30) were conducted in runoff plots (2 m x 1 m) to form soil crust. Crust and noncrust soil samples were collected and subjected to a hydraulic flume (4 m x 0.2 m) to determine the soil detachment capacity (D c). Soil crust traits such as bulk density (BD), clay content (C), crust thickness (CT), macroaggregate content (MA), soil cohesion (Coh), and soil penetration resistance (PR) were measured, and soil erosion resistance (Kr, rill erodibility, and tau c, critical shear stress) was calculated. BD and C exhibited minimal variation, whereas CT, MA, Coh, and PR demonstrated moderate temporal variability across varying rainfall durations. In contrast to the absence of rain, the D c decreased by 50% for R5, 58% for R10, 67% for R15, 71% for R20, and 83% for R30, the K r decreased by 43%-764%, and tau c decreased by 18%-36%. D c exhibited a power function relationship with the hydraulic parameters, with the stream power being the most suitable for capturing the variations in D c. The path analysis indicated that PR and CT were the primary factors directly influencing D c, with standardized path coefficients of -0.38 and -0.48. The variability in D c can be satisfactorily predicted using a triple exponential power function involving the stream power, PR, and CT. This study revealed that variability in soil crust traits significantly influences soil detachment and offers a framework for predicting and managing soil erosion in areas susceptible to soil loss.