Analyses were conducted, using the Early Childhood Longitudinal Study, Kindergarten Class (ECLS-K) database, to compare teachers' judgment of reading skill with direct measures of reading performance for kindergarten, first grade, and third grade students. Teacher judgments of kindergarten students' reading skill significantly predicted first and third grade performance on direct reading measures. In addition, concurrent validity quotients were moderate ranging from r = .58 to .71. These concurrent relationships were further investigated to determine whether socioeconomic status (SES) or teacher experience significantly moderated the concurrent validity estimates. While teacher experience did not significantly moderate the relationship between teacher judgment and direct measures of reading, a small but significant moderation effect was found for SES. That is, SES appeared to differentially bias estimates of reading skill. Judgment bias due to SES level appeared to be greatest when teacher judgments were higher on the rating scale (i.e., proficient) for kindergarten and first grade students. However, for third grade students, judgment bias due to SES was greatest for students lower on the rating scale (i.e., skill not present).