This study investigated the similarity of information provided from a daily behavior report card (DBRC) as rated by the teacher to direct observation data obtained from external observers. In addition, the similarity of ratings was compared with variations of problem severity (mild, severe) and teacher training (none, some). Results suggested a moderate association between teacher perceptions of behavior as measured by DBRC ratings and direct observation conducted by an external observer. In addition, 23-45% of the variance in DBRC ratings was consistent with the direct observation data. Severity of the behavior problem or the inclusion of training was not found to significantly affect the similarity of ratings. In summary, results tentatively suggest that the DBRC may be a viable supplement to direct observation for estimating behavior in applied settings. Limitations, future research directions, and implications are discussed. © 2005 Wiley Periodicals, Inc.Teacher ratings of behavior have become an integral part of the evaluation of children's behavior, with some suggestion that such ratings are the most widely used method for assessing classroom conduct problems (Hoge & Andrews, 1992). The types of teacher ratings range from norm-referenced measures of a broad range of behaviors to educator-made questions about a narrower set of classroom behaviors. The latter, which we refer to as a daily behavior report card (DBRC), serves as the focus of the current investigation. A measure is considered a DBRC if a specified behavior is rated at least daily, and that information is shared with someone other than the rater (Chafouleas, Riley-Tillman, & McDougal, 2002). Although various other terms have been used to describe a DBRC when used as an intervention tool (e.g., good behavior note, home school note, daily report card), using such measures as a way to collect information about student behavior may be considered a type of performance-based behavioral recording. In such a performance-based behavior recording procedure, behaviors are identified and described, and then a Likert-type scale is developed that corresponds to specific dimensions of behavior (Steege, Davin, & Hathaway, 2001).Historically, systematic direct observation has been considered an essential component of behavioral assessment (Alessi, 1998;Hintze, Volpe, & Shapiro, 2002). It has been argued that direct observation, although not without limitations, lends itself to more precise (i.e., reliable and accurate) estimates of behavior because the information is collected as the behavior actually occurs.