Sliced Latin hypercube designs are very useful for computer experiments with qualitative and quantitative factors,multiple experiments,data pooling and cross-validation.However,the presence of highly correlated columns makes it difficult to identify the most important input factors.In this paper,we develop a constructive method for sliced (nearly) orthogonal Latin hypercube designs through cascading Latin hypercube designs.The resulting design preserves zero or low correlations among columns and can be divided into slices of smaller (nearly) orthogonal Latin hypercube designs.Examples are given for illustrating the proposed method.