Records kept by teachers and data collected by trained observers were examined as alternatives for measuring allocated and engaged instructional time in elementary school classes. The teacher ratings of student engagement did not correlate with observer ratings of engagement, but a high correlation between teacher ratings of engagement and student aptitude was found. However, teacher records of allocated instructional time were judged to be acceptable, using observed data as a criterion. In general, the predictive validity of observational records is higher than that for teacher records, while that of engaged time is higher than that for allocated time. (Author)