We develop a method to measure the fairness of school matching mechanisms and apply it to the Boston mechanism used in the Chinese college admissions system. Fairness is measured by mismatch, or the gap between the actual and fair matching outcomes. An individual’s mismatch is then related to his/her behavior toward the first choice in the preference list. We find substantial mismatch and significant demographic differences using the 2003 administrative data from college admissions in China. Risk aversion, information disadvantage, and policy favors can explain the gender differences, rural-urban divide and racial gaps found in mismatch and first-choice behavior, respectively.