Monte Carlo techniques have been used to simulate the stochastic responses of the nonisothermal packed bed reactor. Random variables with specified ensemble means, ensemble standard deviations, and probability distributions (normal, uniform, or autocorrelated) were introduced as inputs and parameters into the model equations, and the equations were solved repeatedly to provide a sample of the outputs. Sample means and sample standard deviations both in the steady and unsteady state were evaluated and the probability distributions of the output temperature and concentration characterized. Confidence limits for a specified confidence coefficient have been determined from which overdesign factors could be computed.