Soil erosion has been recognised as a global threat against the sustainability of natural ecosystem. The work in this thesis has been undertaken to assist in combating this threat, and addresses the soil erosion issues associated with urban construction activities. The Revised Universal Soil Loss Equation (RUSLE) was employed in this research work and the parameters associated with the model were calibrated. This model was chosen for calibration, as it has been proven to be an easy to use tool yet providing reasonable results. Large scale rainfall simulators developed at UWS were used for rainfall simulation at two sites with diverse soil types: dispersive clayey soils at Penrith and highly permeable sandy soil at Somersby (Both in New South Wales, Australia). It is concluded that RUSLE can be successfully used in single storms for erosion prediction. Calibrated values of RUSLE parameters are useful in predicting soil erosion from the construction sites in NSW. It is also identified that in rolled smooth land condition, clayey soils are more erodible than sandy soil. Specific support practices such as short grass strips, gravel bags and silt fences are identified as very effective erosion control measures in reducing soil erosion from 45% to 85%. These results will be very useful in soil erosion prediction planning and conservation management in NSW. / Master of Engineering (Hons)