Implementation of active-learning practices into science, technology, engineering, and mathematics (STEM) courses has been shown to increase student learning and performance, improve retention of material, and reduce achievement gaps among different student populations when compared to lecture-based approaches. To this end, we transformed two large-enrollment introductory geoscience courses (160 to 270 and 60 to 190 students) at a large public university in the midwestern United States into active-learning classrooms, and systematically analyzed student achievement data over a 10-year period. One course is required for majors in geology as well as majors in petroleum and architectural engineering, and also serves as a natural science distribution requirement for bachelor degrees. The other course serves as a natural-science distribution requirement for bachelor degrees. In both courses, there were measurable but not statistically significant (<95% confidence) improvements in overall student performance in the transformed course when compared to all student performances in the untransformed course, as judged by overall decreases in percentage of students in the course earning D or F grade, or withdrawal (%DFW). In the majors course, however, there were measurable decreases in %DFW in female students (-9.5% decrease) and also students from underrepresented minority groups (-5.6% decrease). During the 10-year study, female students remained ~30% of the class, whereas underrepresented minority student enrollment increased from 10.8% to 17.1%, consistent with institutional undergraduate enrollment trends. In the nonmajors course, %DFW for first-time freshman decreased by 7.6%. Our data support previous studies that demonstrate a narrowing in student performance gaps with active-learning practices for at-risk and underrepresented populations.