Acceptance criteria for biopharmaceutical products refer to numerical tolerance limits within which the measured result of a particular attribute of a quality lot should fall. Such limits are usually established based upon clinical experience with the drug product, the capability of the manufacturing process, and the analytical method used to produce the measurement result. For a marketed drug because there is limited clinical experience with commercial lots, the acceptance criteria of a quality attribute rely more heavily on manufacturing history. This potentially could result in unrealistically narrow ranges. In this talk, based on quality by design principles, we present a novel approach to setting up the acceptance criteria. By modeling the theoretical relationship between an attribute of interest and efficacy parameters, we are able to identify a range of the attribute that provides quality assurance to the product. The method is illustrated through an application.