Abstract In the existing literature regarding to the multiple lot-sizing problems with random yield, two practical issues: finite number of setups and holding costs seems to be ignored for study. In this thesis, we shall impose these two factors into the model to derive some managerial insights. Assume that the random yield follows an interrupted geometric distribution and the cost structure is linear, various numerical simulations allow us to obtain the following managerial insights. (1) Improving the yield rate will reduce the production cost. (2)Production periods are consecutive and are implemented as late as possible. (3) The optimal lot sizes will converge as the customers’ demands become large. (4) Numbers of setups will converge. (5) Penalty costs and the yield rates have the same effects over the optimal lot sizes. (6) The higher the inventory costs are, the shorter the production periods will be.