This paper considers an auto parts supplier who receives order release updates from its customers and revises its production plan for future periods on a weekly basis. The inaccuracy of the order releases causes significant costs in the form of premium expedited transportation, production overtime, and excess inventory. This setting provides a rich context for studying order release variance, because the supply chain has adopted a just-in-time (JIT) approach where ideal inventory levels are kept at zero. This leads to a high reliance on order release accuracy in order to manage production quantities. This paper presents an optimization model that extends previous approaches focused on optimizing production plans to the JIT setting. Furthermore, based on real order release information provided to the supplier, two simple adjustment heuristic methods are developed. The performances of these approaches are compared with relying solely on order releases received from the customers. The simple median-based adjustment heuristic performs the best of all the approaches. Implications of the analysis are also discussed.