Given the dearth of high-quality curriculum materials aligned with the new standards (NGSS and CCSS) and low student persistence in STEM fields, we sought to develop and test a STEM curriculum that would improve student knowledge, interest, and emotions. A cluster randomized control trial was conducted to assess the impact of Speedometry, a two-unit STEM curriculum that uses familiar toy cars to explore math and science concepts. A total of 1,615 fourth-grade students across 48 classrooms and 17 schools in an urban district participated in the study. Using a 3-level multilevel model (students nested within teachers and schools) and controls for student and teacher characteristics, we found that Speedometry led to significant increases in student knowledge and positive emotions such as excitement. Speedometry also led to significant decreases in negative emotions such as boredom, frustration, and confusion. The curriculum was implemented with high fidelity as evidenced by classroom observations and teacher self-reports.