The problem of rolling planning and adjusting is considered. By analyzing the essential operation properties of the problem, a 0-1 integer programming model is established with the objective of minimizing the total cutting quantity of rolling plan. A two-stage algorithm based on exchange neighborhood searching is constructed. In the first stage, the initial solution is gained through matching and cross-using. In the second stage, the solution in the first stage is optimized by searching the roller neighborhood and selecting the alternative roller with less cutting quantity. The algorithm is verified by the actual production data, and the results show that the cutting quantity is reduced significantly, and the algorithm and model are feasible and effective.