Student performance at university is a strong determinant of individual decisions and future outcomes, most notably labour opportunities. Although published studies have found gender differences in student performance in response to pressure, little is known about such differences when university students respond to different levels of pressure, resulting from different stakes. Based on field data, this study aims to examine gender differences in student performance in response to different stakes when sitting multiple choice tests, a frequently employed exam format at university. To do so, the introduction of continuous assessment in the evaluation system of a university course allows us to exploit a unique quasi-experimental set up in which the same students take similar tests throughout the course but under different levels of pressure, i.e. facing different stakes. Exploiting individual student data in a panel data framework, we find that male students outperform their female counterparts when under high pressure. However, as the stakes faced decrease, the gender gap shrinks and even reverses in favour of female students at the lowest pressure scenario. We also analyse possible mechanisms responsible for the observed gender gap by studying whether students excel or choke under pressure depending on their gender, and by studying gender differences when omitting questions on multiple choice tests. (c) 2021 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license ( http://creativecommons.org/licenses/by-nc-nd/4.0/ )