Analysis of sensitivity and uncertainty of assessment models for external costs, which is monetarization of environmental impacts, of a commercial fusion plant were performed. The assessments covered the plant's entire life cycle, and adopted the ExternE methodology, which had been used to calculate external costs from other energy sources. Based on the SEAFP study, three different power plant designs were considered. The method developed in ExternE to estimate uncertainty gave very large ranges. A statistical error propagation method was employed for this study. Rather than as a single value, model input parameter values were given as distributions, from which random input sets of data were constructed. The models were then run with these sets, and the ensemble of output results was analysed statistically, yielding estimates of the uncertainty due to variation of the model parameteres. More information of parameter variation is needed for a more realistic estimation of model uncertainty, though. Sensitivity analyses were performed by varying all input parameters in a similar fashion. All model parameters were assumed to have a gaussian distribution with standard deviations of 10% of the mean value. The results pointed out the most essential parameters of the models. The sensitivity analyses are also useful for estimating the most effective ways to reduce the model computed external costs.