Classroom observations increasingly inform high-stakes decisions and research in education, including the allocation of school funding and the evaluation of school-based interventions. However, trends in rater scoring tendencies over time may undermine the reliability of classroom observations. Accordingly, the present investigations, grounded in social psychology research on emotion and judgment, propose that state emotion may constitute a source of psychological bias in raters’ classroom observations. In two studies, employing independent sets of raters and approximately 5,000 videotaped fifth- and sixth-grade classroom interactions, within-rater state positive emotion was associated with favorable ratings of classroom quality using the Classroom Assessment Scoring System (CLASS). Despite various protections enacted to secure reliable and valid observations in the face of rater trends—including professional training, certification testing, and routine calibration meetings—emotional bias still emerged. Study limitations and implications for classroom observation methodology are considered.