Art therapists have been unsettled by the findings of a pragmatic randomised controlled trial testing the addition of group-based art therapy to standard care for people diagnosed with schizophrenia. Arguments that the therapy tested was not that routinely delivered in the UK have been fuelled by the arguably scant descriptions of therapy published to date. To inform the important debate about implications of findings, we provide a comprehensive description of therapy delivered in MATISSE. Drawing on accounts of therapists, their supervisors and participants and study documents, we articulate the three models used to deliver therapy. Described as modified studio, phased group and potentially interactive art therapy, the models were differentiated by structure and the degree of interpersonal and types of therapeutic interaction encouraged. Therapists, it seems, began with their ‘usual’ practice and while remaining true to their ethos, modified that to fit the trial context and participants' needs. Such adaptation is consistent with the principles of pragmatic trials which seek to test interventions in circumstances approximating the ‘real world’. MATISSE provides a piece of the puzzle but a plurality of evidence is needed before ‘calling time’ on the debate about the usefulness of art therapy for people diagnosed with schizophrenia.