Background: Hippocampal volume (HV) reduction is well documented in schizophrenia. However, it is still unclear whether this change is a pre-existing vulnerability factor, a sign of disease progression, a consequence of environmental factors, such as drug use, antipsychotic medication, or malnutrition. The timing of HV changes is not well established, but a lack of macrostructural hippocampal brain abnormalities before disease onset would rather support a neuroprogressive illness model. Aim: To investigate the timing of HV changes in emerging psychosis. Methods: A cross-sectional MRI study of manually traced HVs in 37 individuals with an At Risk Mental State (ARMS) for psychosis, 23 individuals with First-Episode Psychosis (FEP), and 22 Healthy Controls (HC) was performed. We compared left and right HVs corrected for whole brain volume across groups using analysis of covariance (ANCOVA) with gender as a covariate. Sixteen of 37 ARMS individuals developed a psychotic disorder during follow up (ARMS-T). The mean duration of follow up in ARMS was 25.1. months. Results: The overall ANCOVA model comparing left HVs across FEP, ARMS and HC indicated a significant general group effect (p<.05) with largest volumes in ARMS and smallest in FEP. ARMS-T subjects had significantly larger left HVs compared to FE but no HV differences compared to HC (p<0.05). Over all groups, we found an asymmetry between the left and right mean HVs and a strong effect of sex. Discussion: The present study suggests that macrostructural hippocampal abnormalities probably occur in the context of the first psychotic breakdown.