This paper presents preliminary, but statistically significant, findings from a study that compares two methods of measuring instructional effectiveness: global evaluation by experts and systematic observation using the SCRIBE Ob.2 software developed at the University of Texas at Austin. Hierarchical instruction of a performance skill (instrumental music) was the topic of study. Eight evaluators, four from music education and four general education evaluators, ranked six videotaped rehearsal segments in order of effectiveness and provided brief comments regarding the criteria used to rank the segments. Independently the same excerpts were analyzed using the SCRIBE software. The SCRIBE observation instrument was used to record data, and the software allowed for configuration of subjects and behavior categories. Judges from within and outside the field of music education consistently identified exemplary and substandard teachers, but rankings of teachers of intermediate effectiveness were less reliable. The SCRIBE software was a useful and generally manageable means for gathering information on student-teacher interactions. It identified ways in which exemplary and less effective teachers differed to a statistically significant degree. This pilot study holds promise for improved evaluation approaches. (Contains 4 tables, 13 figures, and 23 references.) (SLD)