BACKGROUND: The long-term consequences of discontinuing antipsychotic medication after successful treatment of first-episode psychosis are not well studied. We assess the relation between early maintenance therapy decisions in first-episode psychosis and the subsequent clinical outcome at 10 years. METHODS: This is a 10 year follow-up study, spanning Sept 5, 2003, to Dec 30, 2014, of a randomised, double-blind trial in seven centres in Hong Kong in which 178 patients with first-episode psychosis with full positive symptom resolution after at least 1 year of antipsychotic treatment were given maintenance treatment (n=89; oral quetiapine 400 mg daily) or early treatment discontinuation (n=89; placebo) for 12 months. After the trial, patients received naturalistic treatment. Overall this cohort of patients will have received about 3 years of treatment before entering the follow-up phase of the study: about 2 years of maintenance treatment before study entry and 1 year of treatment in the trial. The primary outcome of this follow-up was the proportion of patients in each group (including those for whom direct follow-up was not available) with good or poor long-term clinical outcomes at 10 years, with poor outcome defined as a composite of persistent psychotic symptoms, a requirement for clozapine treatment, or death by suicide. The randomised trial was registered with ClinicalTrials.gov, number NCT00334035, and the follow-up study was registered with ClinicalTrials.gov, number NCT01926340. FINDINGS: Poor 10 year clinical outcome occurred in 35 (39%) of 89 patients in the discontinuation group and 19 (21%) of 89 patients in the maintenance treatment group (risk ratio 1·84, 95% CI 1·15-2·96; p=0·012). Suicide was the only serious adverse event that occurred in the follow-up phase (four [4%] patients in the early discontinuation group vs two [2%] in the maintenance group). INTERPRETATION: In patients with first-episode psychosis with a full initial response to treatment, medication continuation for at least the first 3 years after starting treatment decreases the risk of relapse and poor long-term clinical outcome. FUNDING: Food and Health Bureau, Research Grants Council of Hong Kong, and AstraZeneca.