Financial aid programmes for students in the United States focus increasingly on academic merit, rather than financial need. There is little empirical evidence, however, on the distributional effects of merit–based aid – who benefits or responds most. We develop a bivariate probit model of the enrolment process estimated using data for a large public university over several years. Results show that merit–based aid increases enrolment for all students, but that financially–able students respond disproportionately, even with academic merit held constant. Thus, increased emphasis on merit in financial aid may exacerbate the trend toward greater income inequality in the US, even among students of equal academic merit.