This paper studies the implementation of quotas in matching markets. In a controlled laboratory environment, we compare the performance of two university admissions procedures that both initially reserve a significant fraction of seats at each university for a special subgroup of students. The first mechanism mimics the sequential procedure currently used by the central clearinghouse for university admissions in Germany. This procedure starts by allocating reserved seats among eligible students and then allocates all remaining seats among those who were not already assigned one of the reserved seats in the first part of the procedure. The second mechanism is based on a modified student-proposing deferred acceptance algorithm in which all seats are allocated simultaneously. In theory, the two mechanisms should lead to similar outcomes. Our experimental results, however, suggest that, relative to the sequential procedure, the simultaneous mechanism significantly improves the match outcomes for the beneficiaries of reserved seats.