AbstractCross-sectional studies suggest that hippocampal volume declines across stages of psychosis. In contrast, longitudinal studies indicate that hippocampal volume is stable in the critical period following illness onset. How can these seemingly disparate sets of findings be resolved? In the present study, we examine two previously unexplored reasons for this discrepancy. First, only specific subregions of the hippocampus may change during the early stage of psychosis. Second, there is diagnostic heterogeneity in the early stage of psychosis and cross-sectional analysis does not permit examination of illness trajectory. Some early stage individuals will have persistent illness leading to a diagnosis of schizophrenia, whereas in others, psychosis will remit. Hippocampal volume may be reduced only in individuals who will ultimately be diagnosed with schizophrenia. We acquired longitudinal structural MRI data from 63 early psychosis and 63 healthy control participants, with up to 4 time points per participant collected over 2 years. Subfield volumes were measured in the anterior and posterior hippocampus using automated segmentation specialized for longitudinal analysis. We observed a volume deficit in early psychosis participants compared to healthy controls that was most pronounced in the anterior hippocampus, but this deficit did not change over 2 years. Importantly, we found that anterior cornu ammonis volume is smaller at baseline in individuals who were diagnosed with schizophrenia at follow-up, but normal in those who maintained a diagnosis of schizophreniform disorder over 2 years. Smaller hippocampal volume is not diagnostic of psychosis, but is instead prognostic of clinical outcome.