An algorithm for the construction of an explicit piecewise linear state feedback approximation to nonlinear constrained receding horizon control is given. It allows such controllers to be implemented via an efficient binary tree search, avoiding real-time optimization. This is of significant benefit in applications that requires low real-time computational complexity or software complexity. The method has a priori guarantee of asymptotic stability with region of attraction being a close inner approximation to the stabilizable set. This is achieved by ensuring that the approximation error does not exceed the stability margin.