There are different approaches for the mixed-model assembly-lines sequencing problem. In this paper the goal of minimizing work overload is treated. This approach considers the existence of time windows in each work stat ion. Different versions of a product are considered to be assembled in the line (e.g. car industry), which require different processing time according to the work required in each work station. Long sequences of rich-work products can lead to produce work overload when stations cannot fulfil all the assigned tasks. Since solve this problem optimally is difficult, we test local search and a hype r- heuristic procedure. A computational experiment is used to detect the performance of the proposed procedures.