The results are provided of an initial analysis of the reliability (generalizability) of the System for Teaching and Learning Assessment and Review (STAR) as a comprehensive measure of classroom teaching and learning for making teacher certification decisions. The STAR contains 140 indicators of teacher effectiveness and student learning, which are classified into four performance dimensions (preparation, planning, evaluation; classroom/behavior management; learning enviornment; enhancement of learning) that are operationalized by 23 teaching and learning components (TLCs). The STAR is used to train principals, master teachers, supervisors, college faculty, and other educators to thoroughly assess beginning and experienced teachers' classroom performances for the purpose of renewable professional certification. Study data were collected during the spring of 1989 in 11 schools in an urban school district in southeast Louisiana. Forty-six teachers were assessed on the STAR on 2 occasions by each of the 3 observers for a total of 276 assessments (46 teachers x 6 observations). A four-facet General Purpose Analysis of Variance System was used with the following factors: teachers, assessor type, assessment indicators, and assessment occasions. A generalizability coefficient (GC) was computed for each of the STAR TLCs. The results suggest common perspectives across assessor types as they view classroom teaching and learning over multiple teachers and multiple lessons. The GC for the STAR thinking skills component was the highest of all GCs. The results generally support the STAR's initial reliability. One figure and four data tables are provided. (RLC)