SummaryEarly intervention in psychosis services produce better clinical outcomes than generic teams and are also cost-effective. Clinical gains made within such services are robust as long as the interventions are actively provided. Longer-term data show that some of these gains are lost when care is transferred back to generic teams. This paper argues that sustaining these early gains requires both a reappraisal of generic services and an understanding of the active ingredients of early intervention, which can be tailored for longer input in cases with poorer outcome trajectories.