The paper presents a methodology for the analysis of just-in-time production systems that are expected to deliver a fixed quantity of parts to its customers on a periodical basis, and whose equipment is submitted to random failure and repair processes. The methodology is based on the determination of the density function of T-H, the cumulated halting time of the production system between deliveries. The paper firstly shows how this function can be obtained from the internal structure of the system and individual equipment failure processes. Then, it shows how the reliability of the deliveries, and several production costs, e.g., extra work, delivery failures penalties, loss of sales profit, and safety stocks, may be assessed using this function. The practical application and usefulness of the methodology is illustrated in the final part of the paper for a typical production system of the automotive industry.