To gain a better understanding of persistent patterns of gender differences in mathematics learning outcomes reported in the literature, this thesis explored the relationship between Victorian Grade 7 students' beliefs about themselves as learners of mathematics and classroom environment factors which might influence those beliefs. The affective variables of interest were those included in Fennema and Peterson's (1985) Autonomous Learning Behavior model. This classroom- based model postulates an explanation for gender differences in mathematics learning outcomes. The study was conducted in two parts. A large scale survey (part 1) established that mathematics teachers who are perceived to be interested in their students as individuals and provide opportunities for interaction (personalisation), who encourage participation in classroom activities, and who emphasise investigative skills are likely to promote functional beliefs in their students. In- depth case studies of two grade 7 mathematics classrooms formed part 2 of the study. Several factors common to both classrooms were identified which provided partial explanations for variations in individual students' beliefs inferred from the self-report data and from classroom observations, differences in beliefs among the targeted students, and gender differences in the beliefs of the male and female cohorts. The classroom climates were found to be male- oriented and seemed to facilitate males' learning over females'. The contextual settings of tasks were frequently male-stereotyped, stereotyped classroom behaviour was tolerated, and competitive activities, particularly traditional formal testing seemed to be more highly valued than cooperative endeavours. The membership composition of small groups seemed to affect students' engagement in and affective reactions to cooperative learning tasks. The teachers appeared to hold higher expectations for males than for females and, at times, to reward students' efforts inappropriately. The findings did not challenge the Autonomous Learning Behavior model. Rather the range of classroom-based factors said to influence students' beliefs about themselves as learners of mathematics has been broadened.