This paper models earnings of male and female Bachelor's graduates in Canada five years after graduation. Using a university fixed-effect approach, the research finds evidence of significant (fixed) variations in earnings among graduates from different universities. Within universities changes over time in various characteristics are correlated with changes in graduates' earnings. Increases in undergraduate enrollment are associated with declines in subsequent earnings for graduates, suggesting crowding out. For men, but not women, increases in the professor/student ratio are associated with meaningful gains in students' subsequent earnings. Models that do not condition on a student's major show increased effects of changes in a university's characteristics, with estimated effects rising up to almost two-fold. For women in particular, changes in several university characteristics are strongly associated with changes in women's choice of major. Changes in university characteristics are not strongly related to the probability of employment five years after graduation.