This paper describes a soil-structure coupling method to simulate blast loading in soil and structure response. For the last decade, simulation of soil behavior under blast loading and its interaction with semi buried structure in soil becomes the focus of computational engineering in civil and mechanical engineering communities. In current design practice, soil-structure interaction analysis often assumes linear elastic properties of the soil and uses small displacement theory. However, there are numerous problems, which require a more advanced approach that account for soil-structure interaction and appropriate constitutive models for soil. In simplified approaches, the effect of soil on structure is considered using spring-dashpot-mass system, and the blast loading is modeled using linearly decaying pressure–time history based on equivalent trinitrotoluene and standoff distance, using ConWep, a computer program based on semi-empirical equations. This strategy is very efficient from a CPU time computing point of view but may not provide accurate results for the dynamic response of the structure, because of its significant limitations, mainly when soil behavior is strongly nonlinear and when the buried charge is close to the structure. In this paper, both soil and explosive are modeled using solid elements with a constitutive material law for soil, and a Jones–Wilkins–Lee equation of state for explosive. One of the problems we have encountered when solving fluid structure interaction problems is the high mesh distortion at the contact interface because of high fluid nodal displacements and velocities. Similar problems have been encountered in soil structure interaction problems. To prevent high mesh distortion for soil, a new coupling algorithm is performed at the soil structure interface for structure loading. The coupling method is commonly used for fluid structure interaction problems in automotive and aerospace industry for fuel sloshing tank, and bird impact problems, but rarely used for soil structure interaction problems, where Lagrangian contact type algorithms are still dominant. Copyright © 2012 John Wiley & Sons, Ltd.