Bullying between students at school can seriously affect students' health and academic outcomes. To date, little is known regarding the extent to which bullying behaviour is clustered within certain schools rather than similarly prevalent across all schools. Additionally, studies of bullying behaviour in schools that do not account for clustering of such behaviour by students within the same school are likely to be underpowered and yield imprecise estimates. This article presents intraclass correlation (ICC) values for bullying victimisation and perpetration measures based on a large representative sample of 106 Australian schools. Results show that bullying is not confined to specific schools and school differences contribute little to explaining students' bullying behaviour. Despite this, seemingly negligible ICC values can substantially affect the sample sizes required to attain sufficiently powered studies, when large numbers of students are sampled per school. Sample size calculations are illustrated. (Contains 5 tables.)