The purpose of this study was to investigate the effectiveness of training and practice in using a self-monitoring technique to enhance the comprehension of intermediate-grade poor comprehenders. Subjects were 129 sixth grade low-average comprehenders from three middle schools, who scored from stanine three to stanine six on the reading comprehension subtest of the California Achievement Test. Subjects' treatment was based on the school they attended. Three comparable middle schools were randomly elected and assigned one of the following treatments: (1) training with practice, (2) training, or (3) neither training nor practice. Training in rating and hypothesis formation was conducted by the investigator over a three- day period. Materials used in training ranged from sentences to passages three to four paragraphs in length. Students were taught to rate pass ages on a 1-2-3 rating scale and to use hypothesis formation to enhance comprehension. Following training, the training with practice group used rating and hypothesis formation for three weeks during their regular reading lessons, under the guidance of the classroom reading teacher. After training and practice, all three groups were tested in two sessions. Three measures were used to test for differences among groups. These were a "global" comprehension test, a rating-with-response test requiring a match between perceived knowledge and demonstrated knowledge, and a test requiring the detection of embedded errors. Data were analyzed using a multivariate analysis of covariance. Scores were adjusted using a covariate of reading ability. No significant differences were found among groups. Implications for research and instruction included: training studies such as this may be most effective if done over an extended period of time in an ecologically valid setting; further study is needed to determine whether the Rating-with-Response and Embedded Error measures employed here give useful information about comprehension processing.