In this study, evaluation of direct behavior rating (DBR) occurred with regard to two primary areas: (a) accuracy of ratings with varied instrumentation (anchoring: proportional or absolute) and procedures (observation length: 5 min, 10 min, or 20 min) and (b) one-week test-retest reliability. Participants viewed video clips of a typical third grade student and then used single-item DBR scales to rate disruptive and academically engaged behavior. Overall, ratings tended to overestimate the actual occurrence of behavior. Although ratings of academic engagement were not affected by duration of the observation, ratings of disruptive behavior were, as the longer the duration, the more the ratings of disruptive behavior were overestimated. In addition, the longer the student was disruptive, the greater the overestimation effect. Results further revealed that anchoring the DBR scale as proportional versus absolute number of minutes did not affect rating accuracy. Finally, test-retest analyses revealed low to moderate consistency across time points for 10-min and 20-min observations, with increased consistency as the number of raters or number of ratings increased (e.g., four 5-min vs. one 20-min). Overall, results contribute to the technical evaluation of DBR as a behavior assessment method and provide preliminary information regarding the influence of duration of an observation period on DBR data. (Contains 6 tables.)