Using field experiments, I investigate if provision of (social) information can increase incoming university students' attendance in a voluntary remedial math course. In Intervention 1, treated students receive an invitation letter with or without information about a past sign-up rate for the course. In Intervention 2, among those who signed up for the course, treated students receive reminder letters including or excluding information on how helpful the course had been evaluated by previous students. On average, no treatment increases participation in the course, but further analyses reveal that the effects in Intervention 1 are heterogeneous along two dimensions: First, suggesting salience as a mechanism, both types of information raise attendance among students who enroll late in their study program, which in turn increases their first-year performance and closes the achievement gap to early enrollees. Second, the effect of the letter with information about the past sign-up rate depends on the predicted ex-ante sign-up probability. Students whose probability falls just short of the past sign-up rate increase sign-up and participation, while the opposite is true for students whose sign-up probability exceeds the social information. Along this dimension, however, the changes in attendance do not carry over to academic achievements.