Abstract: By taking skip trajectory air-to-surface missile as example, the application of modern design of experiment (DOE) technology, namely latinized nearly orthogonal hammersley sequence sampling (LNOHSS), in multidisciplinary design optimization (MDO) was studied. Design experiments were conducted using LNOHSS technology, sampling data was compared with the data of conventional latin hypercube sampling (LHS). Metamodels of exact disciplinary models were built based on support vector machine (SVM). The reliability of metamodel responses was checked by regression analysis of validate data set. The example system was optimized based on metamodels, the correctness of metamodels was validated by comparing optimal results with exact analysis. LNOHSS is a highly effective DOE technology. It provides better orthogonality and uniformity than conventional LHS. LNOHSS technology is of great application value to the MDO of complex systems.