Soil erosion modulus from laboratory modeling is often used to predict soil and water loss for a specific area in the field. Therefore, this study aims to investigate the erosion difference between laboratory and field simulation under various intensities of rainfall and areas on a loessal slope in western Shanxi, China. An emphasis was placed on the laboratory modeling and in-situ simulated rainfall events in the field. A systematic analysis was made on runoff modulus, erosion modulus, sediment discharge of unit width and rill development between laboratory and field. The intensities of simulated rainfall were set as 50, 60, 70, 80, 90, 100, 110, 120 mm/h, combined with natural rainfall events in the study region. A coefficient of uniformity above 85% was, considered in the simulated rainfall, similar to the raindrop distribution and size in the natural rainfall. Calibration of rainfall intensities was conducted at the beginning of each experiment. In the field, the runoff/erosion catchment plots were established in the size of 2, 3, 4, 5 m (length) and 2 m (width) in Wangjiagou small watershed in western Shanxi, while in the laboratory, that in the size of 2, 3, 4, 5 m (length) and 0.5 m (width) in the Taiyuan University of Technology. The soil surface was bare, where the soil type was loessal, and the slope gradient was 20°. Initial water content of soil was determined all the same in simulated experiments. Each rainfall event was repeated two times. The period from the rainfall beginning to runoff occurrence was recorded as "runoff occurrence time" during each rainfall event, where the duration of each rainfall simulation was 30 min from the appearance of runoff. All samples of runoff and sediment were collected in the polyethylene bottles with the volume of 1 L at the bottom end of the plot at 2 min intervals, as the final runoff volume and sediment yield. The erosion modulus and runoff modulus were analyzed with the rainfall intensity and area, in the field and laboratory, indicating significant correlation between rainfall intensity (slope length) and sediment yield. Rill morphology and sediment discharge of unit width were further measured to explore difference between field and laboratory. The results showed that laboratory measurements were greater than those of field in-situ simulation. When the slope area of field was 4 times that of the laboratory, the runoff and sediment yield were not 4 times, where the larger the area was, the smaller the ratio of erosion was, indicating that the amount of soil erosion in the field cannot be predicted simply by the laboratory measurements. The rainfall intensity had also greater impact on the runoff erosion than the area. Under the same rainfall condition, the rill was more likely to occur on the laboratory slope surface, and more developed than on the field, which was more inclined to cut rill deep to enhance the runoff erosion force of laboratory slope. Under certain rainfall intensity and slope length conditions, the mass flux tended to be stable after the first peak at the 10-14 min in the field, while the peak appeared at 4 min in the laboratory, where the value was 1.58-10.40 times of that in the field. It showed that the sediment discharge of unit width and its fluctuation in laboratory were higher than that in the field, and the response time was shorter.