Soil detachment is the initial phase of soil erosion and is of great significance to study in seasonal freeze-thaw regions. In order to elucidate the effects mechanism of freeze-thaw cycles on soil detachment capacity of different soils, a sandy loam, a silt loam, and a clay loam were subjected to 0, 1, 5, 10, 15, and 20 freeze-thaw cycles before they were scoured. The results revealed that with increased freeze-thaw cycles, soil bulk density and water-stable aggregates content decreased after the first few times and then kept nearly stable after about 10 cycles, especially for sandy loam. The shear strength of all soils gradually decreased as freeze-thaw cycles increased, except the values of clay loam increased subsequent to the 5th and 15th cycles. After the 20th cycle, the degree of decline of silt loam was the greatest (77.72%), followed by sandy loam (63.18%) and clay loam (39.77%). The soil organic matter of clay loam was much greater than silt loam and sandy loam and all significantly increased after freeze-thaw. Soil detachment capacity of silt loam and sandy loam was positively correlated with freeze-thaw cycle, which was contrary to findings for clay loam. The values of clay loam increased at first and then decreased during the cycles, reaching minimum values at about the 15–20th cycle. After the 20th cycle, the values of sandy loam and silt loam significantly increased 1.62 and 4.74 times over unfrozen, respectively, which was greater than clay loam (0.53 times). A nonlinear regression analysis indicated that the soil detachment capacity of silt loam could be estimated well by soil properties (R2 = 0.87, p < 0.05). This study can provide references for the study of the soil erosion mechanism in seasonal freeze-thaw regions.