Internships for college students are widely promoted as a "high-impact" practice, yet the academic literature is limited by terminological imprecision, lack of data on intern demographics, and insufficient attention to the impacts of program format on student academic and developmental outcomes. In this mixed-methods study we analyze survey (n=1,129) and focus group (n=57) data from students in three diverse U.S. colleges by using inductive thematic analysis, chi-square, and hierarchical linear modeling to document intern characteristics, access-related problems, program structure, and impacts on student outcomes. Results indicate that internship participation varied significantly by race, institution, enrollment status, and academic program, and that 64% of students who did not take an internship had desired to do so but could not due to scheduling conflicts with work, insufficient pay, and lack of placements in their disciplines. Students also reported high degrees of supervisor support, supervisor mentoring, and relationship between internships and academic programs--all program features that were significant predictors of students' satisfaction with internships and perceived value for their career development. Based on these results, we propose a processual model for studying internships, and we discuss implications for career advisors, faculty, and postsecondary leaders. Specifically, we urge employers, colleges and universities to ensure equitable access to internships for all students, to screen employer hosts for mentoring quality and capacity, and to recognize internships can be simultaneously a positive transformative experience for students and a vehicle for reproducing inequality.