Relief supply inventories are important in disaster response operations. To address the problem of dead stocks and avoid relief supply expiration, such inventories must be renewed periodically. Different from the traditional shelf-life rotation mechanism (i.e. renewal of the entire inventory at the expiration date), which pays more attention to decreasing the quantity of dead stocks, the novel period rotation mechanism that we design in this study will gradually rotate relief supplies according to their shelf life. We propose a mathematical programming model to minimise the total cost of relief organisations and conduct a case study and analyse the sensitivity of key parameters using real-world data. Results show that our rotation mechanism can substantially reduce the financial burden on relief organisations and improve the quality of relief supplies for victims. The novelty of our study includes: (1) We introduce a new approach that can rotate multiple times, through which we consider the quality of relief supplies; (2) We incorporate recycling and remanufacturing into the rotation process; and (3) We adopt an inductive reasoning method to develop the model with demand and time uncertainties. We prove the feasibility of our model and contributions of our study to humanitarian logistics practice and research.