A study was conducted to investigate the effects of high and low relevant text underlining and annotating (making margin notes or other marks to indicate important material) on test performance. The subjects, 67 college freshmen, were randomly assigned to one of four experimenter-generated conditions: (1) high relevant underlining; (2) high relevant underlining and annotating; (3) low relevant underlining; or (4) low relevant underlining and annotating, or to a control condition in which students generated their own text markings. During two sessions, all subjects took a test of prior knowledge, read an assigned passage, and took a 24-item multiple choice test consisting of 12 high and 12 low relevant questions. Data analysis showed that subjects in the high relevant condition answered more high relevant questions correctly while low relevant subjects answered more low relevant questions correctly. Subjects who generated their own underlining did not perform significantly better than did those who were given experimenter-generated underlining. The findings show that researcher underlining and annotating have the effect of directing students' attention and are as effective as student markings. However, experimenter markings place different demands on the reader, greatly reducing the number of his or her decisions as to what is or is not important, thus making the reading task easier. (FL)