Accurate school-level enrolment predictions are critical for local government and planners to allocate education-related resources and ensure necessary school infrastructure is provided. Although individual-level school census data are increasingly available for education departments, there is little evidence that these data are being used to improve school enrolment projections. This paper fills this methodological void by developing a School Transition Estimation and Projection (STEP) modelling framework. The STEP model simultaneously projects all schools in a city or region by all academic levels (e.g., Preschool to Year 12) and across three school sectors (Public, Catholic and Independent) using cohort information on the sources of enrolment change (i.e., Preschool entries, within district transfers, inmigration, out-migration, Year 12 graduates) and their trends over time. In this paper, we present the model and its application to schools in the Australian Capital Territory. We also show how the model greatly increases the capacity to understand and predict enrolment change, while increasing efficiency and reducing bias.