Background: Mounting evidence suggests that childhood trauma is associated with brain alterations and these may be linked to a dysfunction of the stress response system, the Hypothalamic–Pituitary–Adrenal (HPA) axis. However, the relationship between these factors remains unclear. Likewise, it is still uncertain whether the biological correlates of childhood trauma in patients with psychosis differ from those present in individuals with the same exposure but without the illness. Methods: Brain structure was evaluated with a 3T MRI scan in 86 first episode psychosis patients (FEP; 49 positive for moderate/severe childhood abuse) (mean age: 27.8 SD ± 9.1 years) and 64 healthy controls (30 positive for moderate/severe childhood abuse) (mean age 29.1, SD ± 8.6 years). First, a 2-way ANCOVA analysis using FreeSurfer was performed, vertex-by-vertex, to explore between-group differences in cortical thickness related to case or control status, and to exposure or no exposure to childhood abuse. Finally, we investigated the relationship between Cortisol Production During the Day (CPD) and the cortical thickness of regions in which a significant association with abuse was identified. Results: Individuals with moderate/severe childhood abuse, irrespective of the presence of psychosis, showed cortical thinning of the right medial-orbital-frontal gyrus and the lingual gyrus (all P < .001 FWE corrected), and their cortical thickness was negatively correlated with the CPD (r = −.30; P = .003 and r = −.24; P = .02, respectively). Furthermore, FEP with abuse reported thinning of the right cuneus, right latero-orbito-frontal gyrus, right postcentral gyrus, right precentral gyrus, right superior-frontal gyrus and right inferioparietal gyrus, while controls with abuse thickening of the same areas (all P < .001 FWE corrected), suggesting an interaction between group (patient/control) and abuse. Finally, among these regions the right cuneus, right latero-orbito-frontal gyrus, right superiofrontal, and the right inferioparietal gyrus were negatively correlated with CPD in controls (r = −.48; P = .003; r = −.45; P = .005; r = −.59; P < .001; and r = −.43; P = .008 respectively) but not in cases. Conclusion: These results suggest that exposure to childhood abuse has a long-term effect on the adult brain in areas involved in social adjustment, mood control, and drive. Interestingly, this effect is divergent in cases and controls, suggesting a specific vulnerability in individuals who would eventually develop psychosis. Furthermore, we found that the thinner these areas, the higher the concentration of cortisol, potentially pointing to a link between corticosteroid toxicity and the brain. Finally, this relationship was established only in controls resilient to abuse exposure, possibly implying an adaptive mechanism to environmental stress.