The effects of varying the formats of behavior rating scale items on teacher ratings of student hyperactivity were investigated. Two hundred forty-two teachers were asked to rate a variety of children; some had been identified as hyperactive by physicians, parents, or teachers; some were not considered hyperactive; and others were randomly selected. Four rating scales were used: the Behavior and Temperament Survey (BTS), School Behavior Survey (SBS), Pupil Behavior Rating Survey (PBRS), and Abbreviated Symptom Questionnaire (ASQ). The teachers rated five or six children with the BTS, PBRS, or SBS. Two or six weeks later, they rated the same children with two of the following three measures: the BTS, SBS, or ASQ. Test-re-test reliabilities for the two major rating scales, the BTS, and SBS, were examined in detail. The items on these scales varied in format, items were negatively or positively worded. The ASQ and PBRS were included for comparison of concurrent validity. Teachers' responses to items which were worded negatively or positively on different scales were discussed. Data indicating test-retest reliability, concurrent validity, known group validity, and classificatory efficiency were presented, as well as sample items from each scale. It is concluded that reliability and validity of teacher ratings can be increased when the precision of the ratings is improved. (Author/GDC)