This article describes the effects of an analysis-of-practice professional development (PD) program on elementary school students' (Grades 4-6) science outcomes. The study design was a cluster-randomized trial with an analysis sample of 77 schools, 144 teachers and 2,823 students. Forty-two schools were randomly assigned to treatment, (88.5 hours) of integrated analysis-of-practice and content deepening PD (over the course of one year) while 35 schools were randomly assigned to receive an equal number of PD hours in science content deepening alone. Students' content knowledge, as measured by a project-specific test, was compared across treatment groups. The effect size for this comparison was 0.52 standard deviations in favor of students whose teachers participated in the PD that included analysis-of-practice. This effect compares favorably to that of other elementary school interventions whose effectiveness was studied with a narrowly focused outcome measure. Analysis of the demographics of the study schools suggests that the treatment effect could be relevant outside the local study context. Implications for future research include tests of mediation for teacher-level outcomes and efficacy tests of specific teaching strategies (intervention subcomponents).