Awareness of optimal behaviour states of children with profound intellectual disability has been reported in the literature as a potentially useful tool for planning intervention within this population. Some arguments have been raised, however, which question the reliability and validity of previously published work on behaviour state analysis. This article sheds light on the debate by presenting two stages of a study of behaviour state analysis for eight girls with Rett syndrome. The results support Mudford, Hogg, and Roberts' (1997,1999 ) concerns with the pooling of participant data. The results of Stage 2 also suggest, however, that most categories of behaviour state can be reliably distinguished once definitions of behaviours for each state are clearly defined.