This work is a comparison of parametric and nonparametric methods for constructing tolerance limits. It includes a historical review of the applicability of the normality assumption and the development of the concept of tolerance limits. The latter emphasizes the differences between the parametric and the nonparametric methods. The comparison is made on the basis of the results obtained by sampling from three types of distributions-normal, truncated normal, and uniform. The results are inconclusive insofar as it was not possible to make a general statement regarding the relative value of the two methods. The considerations necessary for a user to evaluate the selection of a method are discussed. Recommendations for additional work in the area are included.
