BACKGROUND: People experiencing their first episode of psychosis have high risk of suicide, and programs specializing in early psychosis have not always achieved reduced risk. The present study analyzes patterns of suicide ideation, self-harm, and suicide attempts within the Connection Learning Healthcare System of 23 early psychosis programs in Pennsylvania and Maryland that follow the Coordinated Specialty Care treatment model. METHOD: People with first episode psychosis (n = 1101) were assessed at admission and every six months using a standardized battery that included self-reported past-month ideation and clinician-reported past-six-month ideation, self-harm, and suicide attempts. RESULTS: At admission, there were 28 % rates of self-reported past-month suicide ideation and 52 % rates clinician-reported past-six-month suicide ideation, 23 % rate of clinician-reported self-harm, and 15 % rate of attempts. After the first six months of treatment there were significantly lower rates of clinician-reported suicidality (with reductions of at least 77 %), and after the first year of treatment there was significantly lower self-reported ideation (with approximately 54 % reporting lower past-month ideation). Changes were not accounted for by differential early discharge. A range of psychosocial variables predicted within- and between-subject variability in suicidality. Social and role functioning, depressive symptom severity, and a sense of recovery were significant within-subject predictors of all four measures of suicidality. CONCLUSIONS: Compared to admission, we observed substantially lower rates of suicidality within the first year of treatment for clients with first episode psychosis in Coordinated Specialty Care. Reductions were predicted by some of the variables targeted by the treatment model.