The final year dissertation is an important part of an undergraduate degree which delivers a wide range of subject-specific and transferable skills. It plays a significant part in students' learning development and overall experience of university. Finding the right project is emotionally important to students and may underpin their subsequent motivation and engagement. Little is known, however, about how students make this important choice. This study aimed to learn more about students' experiences of choosing a dissertation, how their choice processes varied and whether their choices worked out well for them. It surveyed 150 undergraduates in natural sciences at a UK university, asking a mix of qualitative and quantitative questions. Findings indicate that students value a range of factors when choosing their dissertation, most prominently interest in the subject and approach but also their existing familiarity with the area, the perceived benefits and demands of the work and staff support. Multivariate analysis suggests a variety of choice processes are in operation, with some students valuing content factors and others trading these off against relational ones. With hindsight, 91 respondents (60.7%) felt their choice process had worked well and 87 (58%) would choose the same way again. A subset, however, had felt unprepared to choose, and some of these were particularly unhappy with the outcome. The implication for learning development is that helping students learn to make conscious and informed choices and making dissertation modules student-centric is likely to significantly improve engagement and learning, especially for the less confident.