Inventory management strategy always plays a very important part in supply chain management. The mathematical models proposed by most researchers consider that the production process to be dependable and assume that inventory would not be defective. Therefore, this paper presents a two-tier integrated production-inventory supply chain model that incorporates a defect rate while accounting for an unreliable production process. If defective items are identified, they will be returned to the upstream vendors for reprocessing, after which the repaired products will be shipped back to the buyers. We designed a linear regression equation which combines the concept of decision support system (DSS) to mitigate the uncertainty in ambiguous situations, and then based on the data of the past years to predict more precisely estimate buyer demand. The main goals of this paper are to minimize the expected joint total cost and to identify the best solution considering the presence of defective items, as illustrated by our proposed model. The predicted results based on numerical examples are provided to decision makers or managers for reference to help them make the right decisions and avoid corporate losses.