An O(n log n) trust region approximation method to solve 0–1 non-linear programming is presented. Optimality conditions and numerical results are reported.