In this paper, an improved integrated approach for solving the production planning and scheduling problem for multi-item multi-resource systems is proposed. Detailed capacity constraints are considered in the lot sizing model, thus guaranteeing feasible solutions. We use a Lagrangian heuristic to generate an optimal production plan for a fixed sequence of operations. A procedure proposing production plans with relaxed capacity constraints is combined with a smoothing procedure which modifies these production plans to satisfy the capacity constraints. Then, a sequence improvement method is performed to feed the Lagrangian heuristic with different sequences. A Tabu search is used to guide the selection of sequences.