This paper studies how a student's ordinal rank in a peer group affects performance and specialisation choices in university. By exploiting data with repeated random assignment of students to teaching sections, we find that a higher rank increases performance and the probability of choosing related follow-up courses and majors. We document two types of dynamic effects. First, earlier ranks are less important than later ranks. Second, responses to rank changes are asymmetric: improvements in rank raise performance, while decreases in rank have no effect. Rank effects partially operate through students' expectations about future grades.