Abstract: With the dataset of reclaimed soils in Xuzhou mine area of China, neural networks coupled with bootstrap aggregation were developed to predict the soil water retention curves from routinely and easily determined soil properties such as texture, bulk density and saturated water content. Results indicated good agreement between observed and predicted values. The root mean squared of residuals for predicted values of water content were reduced by 7.5%－27.0% when compared with predicted values using back-propagation algorithm. Use of the developed neural network models is attractive because of improved accuracy and because it permits a considerable degree of flexibility toward available input data.