In a cluster-randomized study, we investigate the impact of an argument-based approach to teaching science in elementary school on science learning and critical thinking skills. Forty-eight schools participated in the study, with data on 9,963 students across the 2 years of the intervention. Annual standardized tests assessing science content knowledge were used to evaluate the effect of the intervention on science using a hierarchical linear model. To assess critical thinking abilities, scores for the Cornell Critical Thinking instrument were assessed for the 2,353 students in fifth grade using a multilevel model. While no statistically significant gains were found for science content, there was statistically significant evidence that the intervention was associated with an improvement in critical thinking scores (p < 0.05, ES = 0.167). In particular, the strongest gains were for students in the following groups: individual education plan, free and reduced price lunch, and English language learners.