Whether teachers, in the absence of standardized test scores, can adequately identify students who require remedial services or are at risk of failing a literacy test was studied. During the 1987-88 school year 22 fourth-grade teachers were asked to identify which of 530 students of various ethnic backgrounds would score in the top or bottom quartiles of standardized reading and language tests. In a second study, 29 second-grade, 29 fourth-grade, and 25 sixth-grade teachers predicted which students (n=2,070) would score in percentiles 1 to 15, 16 to 35, 36 to 50, and above on standardized reading and mathematics tests. Students took the Iowa Tests of Basic Skills and a reading test. Fourth-grade teachers had similar accuracy rates for minority and white students, but had lower expectations for students who had been retained. The second study demonstrated that the teachers were better at identifying students who were truly at risk, and that the teachers had difficulty identifying students below average but not seriously at risk. Overall, the teachers were more likely to make pessimistic judgments about minority and low socioeconomic students, suggesting that the teachers were not conscious of how non-academic factors biased or mediated their expectations for students. An appendix contains 15 data tables. (SLD)