We take school admission mechanisms to the lab to test whether the widely-used manipulable Boston-mechanism disadvantages students of lower cognitive ability and whether this leads to ability segregation across schools. Results show this is the case: lower ability participants receive lower payoffs and are over-represented at the worst school. Under the strategy-proof Deferred Acceptance mechanism, payoff differences are reduced, and ability distributions across schools harmonized. Hence, we find support for the argument that a strategy-proof mechanisms "levels the playing-field". Finally, we document a trade-off between equity and efficiency in that average payoffs are larger under Boston than under Deferred Acceptance.