The aim of this study is to suggest an alternative to the essay testing method of assessing structural understanding of concepts, which suffers from time constraints and lack of scoring reliability. In this method, students' relationship judgments between concept pairs are examined directly rather than by means of multidimensional scaling analysis (a common alternative to essay testing). The similarity between student and instructor judgments, and the reliability of students' judgments as predictors of structural understanding as measured by essay examinations, are investigated. Data were collected over three separate examinations in each of two experiments. In the first experiment, numerical judgments of the strength of relationships between concepts were obtained from students in three units of undergraduate general psychology. The second experiment demonstrated that restricting relationship judgments to high reliability concept pairs increases the accuracy of essay score predicitions from predictor measures derived from reliability judgments. Reliability and correlation measures are obtained by means of multiple regression analysis. Results indicated that students' relationship judgments do not need to be analyzed through multidimensional scaling in order for them to be useful in evaluating structural understanding, and that this sample correlational approach is a valid and objective alternative to essay testing. (AEF)