The present paper represents a demonstration of how LISREL V can be used to investigate scale invariance (1) across time (its relationship to test-retest reliability), and (2) across groups. Five criteria were established to test scale invariance across time and four criteria were established to test scale invariance across groups. Using the Coopersmith Self-Esteem Inventory for Children, six models were developed to test the above criteria with covariance matrices obtained from the responses of 722 Black, White, and Hispanic elementary students. Results indicated that correlated uniquenesses existed across time and this produced an overestimate of the test-retest reliability. In addition, the construct of self-concept was shown to be invariant across the three ethnic groups. Thus, LISREL procedures appear to provide a useful technique for studying scale invariance both within and between subjects. (Author/PN)