An analytical procedure is presented for determining the mean and standard deviation of job flow-time in a job shop by computing the mean and variance of a random number of random variables. The procedure is applied in the case of a job shop with machines subject to random breakdowns and in the case of a job shop with two classes of jobs. Simulation results validate the analytical procedure but they also show that significant departures from the equal mean processing times assumption have an adverse effect on the accuracy of the analytical results. An expression for the optimal work in process inventory level is also developed for a job shop with machines subject to random breakdowns or with two classes of jobs.