To develop a depression scale that differentiates consistently not only between normal and abnormal samples but also within such samples, items from the MMPI D scale were selected on the basis of consistent relationship to the major dimension or factor underlying the 60 items as determined by a contextual analysis of responses in each of 4 normative samples of 40 Ss (normal and abnormal divided also by sex). The final 30-item scale, called the D-sub(30) scale, showed greatly improved within-group distinctions, particularly among normals, dimensionality coefficients (r-sub(d)) of .97, .95, and .98 being obtained in cross validation in contrast to values of .67, .45, and .87 for the original. Split-half reliability showed improvement despite shorter length, and test-retest estimates in 2 normal samples were .88 and .92. Since part-whole correlations indicated that D-sub(30) scale scores account for the systematic differences in scores on the 60-item scale, the D-sub(30) scale was recommended for general use in lieu of the original. T score norms based on 424 Ss were presented and scale content discussed. (PsycINFO Database Record (c) 2006 APA, all rights reserved)