The editorial by Morrison et al ^{Reference Morrison, Hutton, Shiers and Turkington1} is timely in suggesting we should re-evaluate the way in which antipsychotic medication is used in the treatment of psychosis, particularly for those very early in the course of illness. Since the beginning of the early psychosis reform period, we have consistently advocated for low-dose antipsychotic treatment of first-episode psychosis complemented with comprehensive psychosocial care. More recently we have argued ^{Reference Francey, Nelson, Thompson, Parker, Kerr and Macneil2} that the success of early detection efforts means that young people are being seen much earlier in the development of their symptoms, and this alters the risk–benefit ratio associated with treatments. As proposed by the clinical staging model, ^{Reference McGorry3} there is a strong rationale for beginning treatment with more benign, but evidence-based psychological approaches and reserving pharmacological agents, which despite their efficacy, can have significant adverse effects for (psychological) treatment-resistant cases. Treatment should be proportional to severity and need. Factors that support the call for change in the use of antipsychotic medications include the well-documented metabolic side-effects of most antipsychotic medications, the possibility that some of the structural brain changes seen in psychosis may actually be produced by antipsychotic medications (although the significance of these changes in relation to course and outcome is still unclear), and the widespread non-adherence to prescribed antipsychotic medications suggesting a lack of consumer engagement and dissatisfaction with the treatments offered. The influence of all of these factors is magnified in the case of young people early in their experience of psychotic illness. Finally, the arbitrary threshold of sustained positive symptoms may be an imperfect guide to the timing of antipsychotic medication use in every patient. Some people with subthreshold psychosis (or attenuated psychotic symptoms) may fail to respond to psychosocial treatments as first line and prove to benefit from antipsychotic medications, while a subset of patients with first-episode psychosis with short durations of illness may not require antipsychotic medication. Our research and that of other groups has indicated that antipsychotic medications are not needed as first-line therapy in subthreshold psychosis. We are also attempting to clarify the timing and need for antipsychotic medication in first-episode psychosis by conducting a randomised controlled trial investigating whether intensive psychosocial treatment is sufficient for recovery in a selected low-risk subgroup. It is possible that the results of this study will support a staged approach to the treatment of first-episode psychosis such that medications with significant side-effects are reserved for cases where safer treatments have not led to full remission and recovery. The study will also provide important information about structural brain changes in psychosis and the contribution of antipsychotic medication to these changes. The results of this randomised controlled trial will enhance available information about the risk and benefits of treatments for psychosis and thus improve the capacity of clinicians to support informed decision-making by consumers about their treatment.