[Background] Soil detachment capacity by concentrated flow is an important parameter in many physically-based soil erosion models. The seasonal variations in soil detachment capacity and its influencing factors are crucial for accurate prediction of soil loss but not fully quantified in grasslands. [Methods] This study was conducted to investigate the seasonal variation and quantify the potential factors causing changes in soil detachment capacity. The temporal variations in soil detachment capacity for two grasslands and one bare soil were monitored at 20-day intervals from April to October 2011. Undisturbed topsoil samples were taken and subjected to 6 different flow shear stresses (=6.5 -23.4Pa) in a laboratory flume with the fixed bed to measure soil detachment capacity by concentrated flow.Soil bulk density, soil cohesion, and root mass density were measured at the time of sampling to reveal the observed variations in soil detachment capacity. [Results] Distinctive seasonal variation patterns were found throughout the growing season. The seasonal variability in soil detachment capacity under grasslands differed significantly from that of bare soil. Soil detachment capacity in grasslands was relatively high from mid-April to last-June, then declined quickly with grass growth, and maintained at a very low level from mid-July to early-October. In the end of growing season, soil detachment capacity in grasslands increased slightly. For the bare soil, soil detachment capacity was relatively high in the mid- April and declined gradually in May, June, and July. It reached the minimum at early-August, then increased again and continued to the end of the measurement period. The mean soil detachment capacity of bare soil was 10.6 and 23.7 times greater than of switchgrass (Panicum virgatum) and smooth bromegrass (Bromus inermis Leyss) soils, implying that the bare soil was much more erodible than grasslands. Seasonal variation in soil detachment capacity of grasslands could be mainly explained by variations in soil cohesion and root mass density. Soil detachment capacity declined exponentially as root mass density and soil cohesion increased. No statistical significant relationship was found between soil detachment capacity and soil bulk density. [Conclusions] The experimental results showed that soil cohesion and root mass density affected largely on seasonal variation in grasslands, and the soil detachment capacity of grasslands could be well estimated using shear stress, root mass density, root diameter, and soil cohesion. The results of this study aid understanding of soil erosion mechanisms and development of process-based erosion models to simulate the seasonal variation in soil detachment, and provide scientific evidences for the configuration of soil and water conservation measures in small watershed.