Forest litter layer plays an important role in controlling water and soil loss; however, few studies have examined how litter cover affects hydrological processes. In this research, the effects of forest litter cover on interception, runoff, infiltration and soil erosion were investigated using rainfall simulation on a loess hillslope in China. Field experimental plots (10 m × 1.5 m, 5°) with two litter species (needle-leaf species and broad-leaf species) and three litter masses (300, 500 and 800 g m−2) were exposed to two rainfall intensities (30 and 60 mm h−1) for 40 min. A modified Merriam interception model as well as a WEPP (Water Erosion Prediction Project) Hillslope model were subsequently employed to analyze the experimental data of litter interception as well as runoff generation and soil erosion, respectively. Results showed that litter interception storage capacity ranged from 0.55 to 2.10 mm; thus, 1.8–9.2% of total rainfall was intercepted by the litter layer. Interception storage increased with increasing litter mass and rainfall intensity, and broad-leaf litter could intercept more rainwater than needle-leaf litter. Litter cover reduced runoff rates from plots covered with needle-leaf and broad-leaf litter by 18.6% and 25.9%, respectively, compared to the bare plot. Although runoff rates were slightly lower in litter-covered plots than in the bare plot, soil loss was effectively controlled when litter mass levels reached a threshold value of 500 g m−2. Compared to that from the bare plot, the mean soil loss rate for litter-covered plots was lowered by 58.5%, 74.5% and 78.3% for litter masses of 300, 500 and 800 g m−2, respectively. These results indicate that the litter layer was more conducive to controlling soil loss than runoff generation. Furthermore, the determination coefficient and Nash-Sutcliffe efficiency coefficient of the modified Merriam model ranged between 0.838–0.998 and 0.912–0.998, respectively, indicating that this model could successfully simulate the litter interception process. The WEPP model predictions also agreed well with the measured runoff and soil loss rates, but may have under-estimated the measured data in low rainfall intensity events and over-estimated the measured data in high rainfall intensity events. The performance of the WEPP model could be obviously improved by reducing the effective hydraulic conductivity and soil erodibility. Overall, this study will enable the development of more accurate modeling approaches and lead to a better understanding of hydrological processes under litter cover conditions.