This study used data from the Assessment Tools for Teaching and Learning project, which involved data on the academic performance of more than 90,000 New Zealand students in six subjects (i.e. reading, writing and mathematics in two languages). Two sub-samples of this dataset were included for detailed re-analysis to test the general applicability of the Australian model of Quality of School Life (Ainley et al. 1986) in the New Zealand context. The first sample comprised 336 year 8 students from elementary schools and the second sample consisted of 272 year 10 students from high schools. Furthermore, two structural equation models were developed and tested, expressing relationships between students' quality of school life perceptions, students' attitudes to mathematics, and their effects on mathematics achievement. The quality of school life questionnaires scales [Ainley and Bourke, in Res Pap Educ 7(2):107-128, 1992] were used as indicators of students' perceptions regarding learning, teachers and peer relationships. The model proposed that perceived quality of school life would affect students' attitudes of liking and confidence in mathematics, which would in turn affect their academic performance. After controlling for other variables in the model, students' perception about their self-efficacy to learn mathematics was more directly related to outcomes than to perceptions of teacher quality or peer involvement. Data analyses revealed no apparent relationships of these factors to mathematics achievement. Moreover, results for both samples led to the conclusion that the perceived quality of learning is connected with "confidence in" and "liking mathematics", which in turn predict students' mathematics achievement.