This article reports a systematic review of research on science programs in grades 6–12. Twenty‐one studies met inclusion criteria including use of randomized or quasi‐experimental assignment to conditions, measures that assess content emphasized equally in experimental and control groups, and a duration of at least 12 weeks. Programs fell into four categories. Instructional process programs (ES = +0.17) and technology programs (ES = +0.47) had positive sample‐size weighted mean effect sizes, while use of science kits (ES = −0.02) and innovative textbooks (ES = +0.10) had much lower effects. Outcomes support the use of programs with a strong focus on professional development, technology, and support for teaching, rather than materials‐focused innovations. © 2016 Wiley Periodicals, Inc. J Res Sci Teach 54: 58–81, 2017