Prompted by the need to develop improved tools for describing and assessing teachers' skills in teaching reading, a study was undertaken to examine the effectiveness of a system for classifying and tallying teacher-pupil verbal interaction during classroom instruction. Specifically, the study investigated the extent of agreement among college students using a time-line format when ranking categories of verbal interaction in three criterion reading lessons, and it examined the use of selected ratios of verbal interaction as a means of discriminating among criterion reading lessons as compared to the judgments of a jury of reading teachers who ranked the quality of the lessons without reference to the verbal interaction factors. Twenty-five graduate students were trained in the Flanders Interaction Analysis technique using the time-line format. The students then coded the verbal interaction of three videotaped reading lessons. These lessons were also viewed by seven reading teachers, who ranked them according to overall effectiveness. The results indicated that the students reached high agreement in tallying the verbal interaction. In addition, the use of the ratio of teacher questions/pupil responses to teacher lecture/pupil initiated comments was found to rank the lessons in the same quality order as did the jury of reading teachers. (Author/FL)