Over the course of one year, we systematically observed instruction in nearly all large gateway STEM courses at the University of California, Irvine to assess the prevalence of promising instructional practices and their implications for student success. More than half of the courses included promising instructional practices. Our most conservative student fixed-effects models suggest that students earn slightly higher grades in courses where instructors use explicit epistemological instruction, frequent assessment, and interactive instruction. Although we find no evidence to suggest that these strategies have lasting effects for the average UC Irvine student, we do find they have unique positive effects on the achievement of first-generation college students.