The purpose of this study was to investigate the differences among schools using student responses in the Trends in Mathematics and Science Study–1999 (TIMSS‐99) data. Schools were classified into two groups based on eighth grade students' overall achievement in science. Two different discriminant function analyses were performed to distinguish high‐ and low‐performing schools, based both on classroom practices and factor structures (student‐centered activities, teacher‐centered activities, attitudes toward technology use, socioeconomic status [SES], and doing well in science). The results indicate that there were significant differences between the two classifications of schools on ten variables regarding classroom practices and activities, and 29 variables regarding classroom practices, attitudes toward science, use of the computer and overhead projector (OHP), parental background characteristics, and need to do well in science. Contrary to general expectations, technology use (computer, OHP, etc.) was found to be negatively related to science achievement. Teachers should be trained on how to use technology in their classrooms. Turkey recently revised its curriculum to a student‐centered approach and this might increase students' ability to transfer knowledge into real life. Teachers and schools should pay more attention to SES effects. Teachers should also work toward building students' confidence in science. © 2007 Wiley Periodicals, Inc. J Res Sci Teach 44: 1417–1435, 2007