Scores on the Myers-Briggs Type Indicator are sometimes found to have bimodal or multimodal frequency distributions in some groups to which the inventory is administered. This report describes a statistical study designed to determine whether the multimodalities in a data set might be attributable to sample fluctuation. A significance test for unimodality in frequency distributions was devised and applied to the data for a large sample of undergraduates. The null hypothesis that the distributions of the four scales on the inventory were unimodal could not be rejected.