During the processes of development and construction of the Shenfu coalfield, generated were large tracts of disturbed land, and large volumes of translocated soil and waste slag, which, unique in soil constitution and complex in underlying surface layer, have become the major source of serious surface runoff and soil and water loss. An artificially simulated rainfall experiment was carried out to explore effects of rainfall intensity and soil fractal dimension on runoff and sediment yielding from undisturbed land, disturbed land, and piles of translocated soil or waste slag, with undisturbed land as CK. Detailed investigation found that the piles of translocated soils and waste slag had slopes at 40°, their natural angle of repose and the disturbed and undisturbed lands had slopes at 18°. Plots in the experiment field were designed to be 3m × 1m, each, and had two sections, each, set up for monitoring of water flow. Rainfall intensity in the experiment was designed to vary from 1.0 ~ 3.0mm min^-1, forming five levels with a interval of 0.5mm min^-1 between every two levels. Before the experiment, the rainfall intensity was calibrated repeatedly until uniformity coefficient of the rainfalls reached 85% or higher. Soil particle compositions of the plots were determined with the pipette method. During every rainfall, flow velocity was measured with the dye tracing method and width and depth of the flow with a thin steel rule. For the first 3 minutes of runoff, samples of runoff and sediment were gathered once a minute, and after that once every 3 minutes. Results show that (1) on piles of translocated soil and waste slag, runoff followed the process of spurting-declining-leveling off, while on undisturbed and disturbed lands, it did the process of rising – leveling off. Runoff rate on the underlying surfaces increased with increasing rainfall intensity. (2) Fractal dimension of soil particles on the underlying surfaces of the four plots displayed an order of D1 (abandoned residue) < D2 (abandoned soil) < D3 (disturbed ground) < D4 (undisturbed ground). Volume of runoff caused by a single rainfall event was found to be in a significant linear relationship with rainfall intensity and a significant power function relationship with fractal dimension. Moreover, between D1 and D2 and between D3 and D4 existed two critical fractal dimensions, i.e. 2.229 and 2.479, which can be used to distinguish types of underlying surfaces. (3) Erosion on the pile of waste slag followed the process of fluctuating – leveling off, while on the pile of translocated soil, it was characterized by multiple peaks and valleys. On undisturbed and disturbed lands, erosion increased in rate first and then gradually leveled off under rainfall 1.0 ~ 2.5 mm min – 1 in intensity, and fluctuated drastically under rainfall 3.0 mm min-1 in intensity Erosion rates on all the four underlying surfaces increased with increasing rainfall intensity. (4) Amount of erosion was found to be in a significant power function relationship with rainfall intensity, and with fractal dimension of soil particles, too. (5) Volume of runoff was in significant linear relationships with rainfall intensity and fractal dimension of soil particles (Mw=-147.43Di+123.46Pi 268.96, R²=0.952), while yield of sediment was in significant exponential function relationships with the two (lnM_s=-11.32D_i 1.32P_i 25.83,R²=0.844). The above findings indicate that disturbed land and piles of translocated soil and waste slag, all formed as a result of mining, differ sharply from undisturbed land in rules of runoff and sediment yield, and underlying surface is an affecting factor of runoff and sediment yield that should not be ignored. Fractal dimension of soil particles, as a quantitative indicator of underlying surface, can be used to effectively predict runoff volume and sediment yield. Obviously this study has some important scientific significance to the establishment of a soil erosion model for mining areas.