'Ihe argument that a normal distribution of a performance indicator may be an inadequate approximation of the true distribution is examined in this paper. Simulated normal and beta distributions of net return for 1,181 swine operations are compared using a Kolmogorov-Simirnov test and stochastic dominance techniques. Results show that for this data set the differences between the simulations are not statistically significant nor does stochastic dominance switching take place. Although the normal distribution represents an acceptable approximation, the added flexibil­ity of the beta distribution may prove more useful in future research.