A study utilized data from four successive school years to explore the stability and correlates of teacher effects in reading. A "value-added" model was used to isolate the effect of the teacher from child demographic variables such as race, poverty, gender, family composition, and special learning needs. Teacher effects in second grade reading were found to have moderate stability over two consecutive years. Estimates of teacher effect stability increased substantially when value-added effects were aggregated over three or more years. Teacher effects in second grade reading correlated dependably with several facets of direct instruction philosophy and practice on a self-report survey. Teachers who demonstrated the highest value-added tended to disagree with the statement"reading and writing develop naturally, like speaking." They endorsed more use of small group instruction and more use of guided practice. Teachers identified as "exceptional" through value-added analysis endorsed more teacher directed activities, more development of word attack strategies and more use of individual student oral reading. Use of systematic motivational strategies and some form of test preparation activity were also endorsed to a greater extent by teachers with high value-added estimates. These findings are consistent with National Research Council findings on prevention of early reading difficulties. A balanced reading approach which utilizes explicit reading skill instruction was associated with higher reading success in second grade classrooms in this study. (Contains 61 references, 5 notes, and 20 tables of data.) (RS)