This thesis examines pupils' attitudes to science as they move from primary to secondary school, and seeks to identify any factors that might influence these attitudes. A review of the literature on attitudes to science research finds that existing studies with similar aims have tended to use cross-sectional samples, many of which were small or unrepresentative, and furthermore that there has often been inadequate control of potentially influential variables. The present work employs a longitudinal study of over 3000 children, between 10 and 12 years old, from schools in Essex. Data has been collected by means of questionnaires, supplemented by a free response section, and by interviews. Information was also collected from over 300 primary and secondary teachers by means of questionnaires, supplemented by a free response section, and interviews. Information was collected, by interviews and from statistical data in the public domain, about all participating schools. The integrated data from the children, their teachers and their schools has been analysed in three different ways: the quantitative data was subjected to a variety of statistical techniques to compare the two sets of data from primary and secondary school as two cross-sections, and to investigate changes for individual pupils taking a longitudinal approach. The qualitative data was subjected to textual analysis and it was also integrated with the quantitative data. These analyses yield conclusions, which inform pedagogy, school management, teacher training, and social justice.