We present primal-dual interior-point algorithms with polynomial iteration bounds to nd approximate solutions of semidenite programming problems. Our algorithms achieve the current best iteration bounds and, in every iteration of our algorithms, primal and dual objective values are strictly improved.