A simple approach was described to incorporate the Monte Carlo simulation technique into a fuel cycle cost estimate. As a case study, the once-through and recycle fuel cycle options were tested with some alternatives (ie, the change of distribution type for input parameters), and the simulation results were compared with the values calculated by a deterministic method. A three-estimate approach was used for converting cost inputs into the statistical parameters of assumed probabilistic distributions. It was indicated that the Monte Carlo simulation by a Latin Hypercube Sampling technique and subsequent sensitivity analyses were useful for examining uncertainty propagation of fuel cycle costs, and could more efficiently provide information to decisions makers than a deterministic method. It was shown from the change of distribution types of input parameters that the values calculated by the deterministic method were set around a 40(th)similar to 50(th) percentile of the output distribution function calculated by probabilistic simulation. Assuming lognormal distribution of inputs, however, the values calculated by the deterministic method were set around an 85(th) percentile of the output distribution function calculated by probabilistic simulation. It was also indicated from the results of the sensitivity analysis that the front-end components were generally more sensitive than the back-end components, of which the uranium purchase cost was the most important factor of all. It showed, also, that the discount rate made many contributions to the fuel cycle cost, showing the rank of third or fifth of all components. The results of this study could be useful in applications to another options, such as the DUPIC (Direct Use of PWR spent Fuel In CANDU reactors) cycle with high cost uncertainty. (C) 1998 Elsevier Science Ltd.