Inventory managers do not predominantly follow normative optimisation models. At best, they introduce a level of bounded rationality in their inventory replenishment decisions. This paper examines the behaviour of inventory decision-makers under continuous review in a decentralised supply chain, using an experimental approach with unknown market demand and local information availability. The analysis reveals that not only the magnitude and the variability of order quantity tend to be larger, but also that the order-time intervals is lengthen and highly variable while moving upstream along the supply chain. The role of the inventory managers' replenishment decisions on the echelon holding, backorder, and total costs, is also investigated. Finally, a normative model is designed and its solutions are compared to the experimental results. It is observed that humans do not operate in a perfectly optimal way, but are generally reluctant to risk increasing backorder costs and reducing inventory carrying cost, even if this would lead to lower total cost.