Background: Discontinuation of treatment in people with first episode psychosis (FEP) is common, but the extent to which this is related to specific adverse effects of antipsychotic medications is unclear. Objectives: To investigate whether antipsychotic discontinuation is associated with the prescription of particular antipsychotics and particular adverse effects. Design: Retrospective cohort study. Methods: We assembled de-identified electronic health record (EHR) data from 2309 adults with FEP who received care from the South London and Maudsley NHS Foundation Trust between 1st April 2008 and 31st March 2019. Associations between antipsychotic medications, clinician-recorded side effects and treatment discontinuation were investigated across a mean follow-up period of 34.2 months using Cox regression. Results: The mean age of patients was 26.7 years and 1492 (64.6%) were male. Among first prescribed antipsychotic medications, discontinuation occurred earlier with haloperidol [hazard ratio (HR) = 2.78, 95% CI = 1.69–4.60] and quetiapine (HR = 1.43, 95% CI = 1.16–1.80) than with olanzapine. Discontinuation occurred sooner when there was evidence of extrapyramidal symptoms (HR = 1.33, 95% CI = 1.08–1.64) or sexual dysfunction (HR = 1.59, 95% CI = 1.03–2.46). Among antipsychotics prescribed at any point during treatment, lurasidone (HR = 1.40, 95% CI = 1.10–1.78) and aripiprazole (HR = 1.09, 95% CI = 1.01–1.19) were associated with earlier discontinuation than olanzapine. Conversely, clozapine (HR = 0.55, 95% CI = 0.41–0.73) and paliperidone 1-monthly (PP1M) long-acting injectable (HR = 0.80, 95% CI = 0.68–0.94) were associated with later discontinuation. Unexpectedly, for antipsychotics prescribed at any stage of treatment, sedation (HR = 0.89, 95% CI = 0.81–0.97), weight gain (HR = 0.73, 95% CI = 0.64–0.83), and multiple side effects (HR = 0.83, 95% CI = 0.76–0.90) were associated with later discontinuation. Conclusion: Earlier treatment discontinuation associated with sexual or extrapyramidal side effects could be related to their rapid onset and poor tolerability. Later treatment discontinuation associated with clozapine and PP1M could be related to the relative efficacy of these treatments. These findings merit consideration when selecting antipsychotic therapy for people with FEP.