Validating and generalizing from holistic observation protocols of classroom practice have proven difficult. These tools miss crucial classroom characteristics, like the type of instruction, the organization of learners, and the level of cognitive engagement that occur differentially in the time span of a lesson. As a result, this study examined the potential of interaction analysis for drawing detailed inferences about science classrooms. Holistic and interaction analysis techniques were applied to an existing set of observational data from a group of high school teachers ("N" = 21) who were participating in long-term professional development. Results indicate that interaction analysis provides crucial details that are otherwise absent. Questions are raised about the overall benefit and use of holistic observation protocols without supplemental information. (Contains 3 tables and 3 figures.)