In retail stores, we often see that products are placed on sales promotion programs in a cyclical manner. Demand for the product increases during promotional period and returns to the regular level at the end of promotion. This cycle is then repeated with almost regular frequency. We refer to this demand pattern as bi-level demand and the supply chains operating in such environments are the main focus of this study. We study the problem of formulating an inventory policy in such supply chain. Because of the periodic steps in the demand function and the uncertainly of the demand, the problem is very challenging and the optimal solution is hard to obtain. First we consider only the retailer, and narrow down focus to the development of the inventory policy for such retailer operating in bi-level deterministic demand environment. We define bi-level demand problem (BDP) as finding the order quantity (how much) and ordering time (when) given the holding and ordering costs. We develop three heuristics for the BDP and a procedure for finding an optimal solution for a subset of BDP. We generated a large set of test problems using appropriate experimental design based on problem parameters and compare the performance of heuristics against lower bounds or optimal solution as the case may be. The heuristics are found to be efficient and provide close to optimal results in most cases. Next, we model a generic version of supply chain consisting of three members—the retailer, distributor and manufacturer—and focus on the problem of formulation of inventory policy for members of such supply chain operating in stochastic (uncertain) bi-level demand environment. Considering the tradeoff between ordering/setup cost, holding cost, and shortage cost, we develop three ordering policies—Moving Average Policy (simple policy), Target Inventory Level Policy (sophisticated policy), and Complete Cooperation Policy (CCP). First two policies assume supply chain members do not share any information and order independently to minimize their costs, while in the last policy members cooperate to minimize total supply chain costs. We evaluate the performance of these policies by testing them on large set of problems using simulation. Results reinforce that cooperation results in lower supply chain costs. We also found that the major cost savings can be achieved by using more sophisticated policies than simple policies like moving average even if cooperation may not be possible. In addition to the problem of formulating inventory policies, we also develop a game called ‘Cola-Game’ which instantiates the supply chain discussed above. The game involves simulating a supply chain and could be played in either independence or cooperation modes. This game has been field tested in engineering and business classes at the University of Toledo. We found that players developed an appreciation for fluctuating demand and its impact on the costs and performance of a supply chain. They also learned the benefits and a monetary evaluation approach for cooperation. Our statistical analysis revealed that as the game progressed, the performance of the teams improved. Thus this game can be used as a tool to educate students and managers on the various issues in supply chain inventory management