This study explored characteristics and variables differentiating more effective and less effective teachers of mildly retarded (EMR) and learning disabled (LD) students. Twenty-four elementary school teachers were designated as more effective or less effective based on students' mean achievement gain scores on the California Achievement Test Educational and personal variables were gathered through information sheets completed by teachers. Teacher behaviors were observed using the Classroom Activity Recording Form (CARF). The 16 behaviors of the CARF are organized into three major headings: Teacher-Directed Instruction (TDI), Independent Work (IND), and Non-Instructional Activities (NIA). Effective teachers could not be identified on the basis of personal or educational vanables, but could be differentiated on teacher behaviors in the classroom. Effective teachers conducted activides in the classroom; made frequent use of teacher questioning; limited independent seatwork, especially silent reading; and allowed time for a certain amount of social interaction with the students. Results are consistent with previous process-product research investigating teacher effectiveness .