The ability of 50 lower middle-class and 25 upper middle-class prereading children to discriminate between pairs of uppercase alphabet letters was tested. A set of 3x5 cards with a sample stimulus in the upper center section of each card and two alternative choice stimuli just below and to the right and left of the sample was used. The 650 total cards were divided into five sets of 130 cards. The two major subject groupings were divided into five groups each and were tested with one set of 130 cards. An analysis of variance showed that the differences between groups both within and across each major grouping were not significant at the .05 level. The range of errors per subject was from 0 to 17 on 130 items. The letter pairs significantly confused were M-N (8); M-W and S-P (5); H-A, I-J, L-J, and K-X (3); and B-X, H-X, N-X, I-L, P-R, A-X, and H-W (2). The uppercase letters with the lightest percentage of errors were in descending order: M, N, K, X, H, P, W, and I. Tables and references are included. (BS)