In this article, the authors describe how classroom observations were used to characterize differences in instructional practices among undergraduate science courses. Student and faculty behaviors and dialogues were quantified using the Teaching Dimensions Observation Protocol (TDOP) in introductory science courses across multiple disciplines (e.g., biology, chemistry, and computer science) at a primarily undergraduate institution. Courses that had been revised to incorporate active learning pedagogies showed significant differences in classroom behaviors compared to unrevised courses. Revised courses had more class time devoted to student-centered behaviors and dialogues such as problem solving, peer interaction, and small group work, and spent less time with students listening to the instructor. Revised courses were also shown to have a "student-centered" instructional profile while unrevised courses used a more instructor-centered "interactive lecture" format. These results show that even when using a relatively small sample size, an observational assessment such as the TDOP is an effective technique to detect instructional differences occurring within introductory undergraduate science classrooms.