This study documents the singing self-efficacy beliefs of 165 primary teacher education students at a New Zealand university. Quantitative and qualitative methodologies were used in order to establish a singing self-efficacy profile for the group under investigation, and to explore the factors which influenced the development of their beliefs. Participants completed a questionnaire containing Likert-scale items and open-ended questions. Data on gender, age and ethnicity were also collected. In-depth interviews were conducted with three students who identified as having low self-efficacy. Analysis of the quantitative data revealed a wide range of singing self-efficacy beliefs. Although the study found no differences in self-efficacy on the basis of gender or age, Maori students appeared to hold more positive beliefs about their singing capability than European/Pakeha students. Three factors, other people, the self, and musical experiences, emerged as significant in the development of students' singing self-efficacy. Because of the prevalence of performance-oriented goal structures in New Zealand school music programmes, and because of fixed rather than incremental conceptions of singing ability, individuals who learn to sing accurately early in their lives are more likely to develop high self-efficacy in singing. The study revealed little evidence of the impact of vocal development research on beliefs about singing capability. Student teachers with low self-efficacy in singing often express anxiety about how they will meet the requirement to teach singing as part of the classroom programme. There is also a strong likelihood that teachers will perpetuate the practices which contributed to their own beliefs about singing capability. It is important therefore to understand how singing self-efficacy develops and to identify classroom goal structures which facilitate the development of both singing skills and positive singing self-efficacy.