In this study, we examined the relationship of special education teachers' performance on the RESET Explicit Instruction observation protocol with student growth on academic measures. Special education teachers provided video recorded observations of three instructional lessons along with data from standardized, curriculum-based academic measures at the beginning, middle and end of the school year for the students in the instructional group. Teachers' lessons were evaluated by external, trained raters. Data were analyzed using many-faceted Rasch measurement (MFRM), correlation and multiple regression. Teacher performance on the overall protocol did not account for statistically significant variance in student growth beyond that of students' beginning of the year academic performance. Teacher performance on an abbreviated protocol comprised of items that had average or higher item difficulties on the MFRM analysis accounted for an additional 4.5% of variance beyond that of beginning of the year student performance. Implications for further research are discussed. [This paper will be published in the "Journal of Learning Disabilities."]