Capstone projects are common in undergraduate programmes, providing students with a culminating educational experience designed to draw on the knowledge and skills accumulated over the course of their studies. While there are many benefits to capstone projects, they are not without challenges. In particular, when these projects are conducted in groups, forming groups to optimise the learning outcomes and managing group dynamics can be challenging. In this article, we report on the analysis of data collected from 346 undergraduate business students who completed capstone projects at a Hong Kong university. Measures included students' learning goal (mastery and performance), satisfaction with their supervisor and group diversity in relation to gender, prior academic achievement, self-report nationality and programme of study. Analysis of this data in conjunction with student grades for the project was conducted to inform improvements in design and delivery of the capstone subject to improve students' learning outcomes. The results showed that for groups consisting of three students, group diversity in respect to prior academic achievement as measured by grade point average (GPA) is positively related to the grade achieved in the capstone project. However, diversity in respect to the nationalities in the group was related to poorer performance. Furthermore, the more teacher-focused the group supervisor's approach was, the worse the grade achieved for the project. The results suggest that groups made up of students of different nationalities tend to have lower grades compared to homogeneous groups. In contrast, having a group with a mix of GPAs can result in higher grades on the project. While these findings have informed our understanding of group performance on capstone projects, work is needed to fully understand what underlies the diversity effects identified which will be explored with future cohorts. Copyright © Routledge.