Disrupted thalamo-cortical connectivity is regarded as a core psychopathology in patients diagnosed with schizophrenia. However, whether the thalamo-cortical white matter connectivity is disrupted before the onset of psychosis is still unknown. To determine this gap in knowledge, the strength of thalamo-cortical white matter anatomical connectivity in subjects at clinical-high risk for psychosis (CHR) was compared to that of first-episode psychosis (FEP) and healthy controls. A total of 37 CHR, 21 FEP, and 37 matched healthy controls underwent diffusion-weighted magnetic resonance imaging to examine the number of probabilistic tractography "counts" representing thalamo-cortical white matter connectivity. We also investigated the relationship with psychopathology. For FEP, the connectivity between the thalamus and parietal cortex was significantly increased (F = 5.65, P < .05) compared to that of healthy controls. However, the connectivity between thalamus and orbitofrontal cortex was significantly reduced compared to both healthy controls (F = 11.86, P < .005) and CHR (F = 6.63, P < .05). Interestingly, CHR exhibited a similar pattern as FEP, albeit with slightly reduced magnitude. Compared to healthy controls, there was a significant decrease (F = 4.16, P < .05) in CHR thalamo-orbitofrontal connectivity. Also, the strength of the thalamo-orbitofrontal connectivity was correlated with the Global Assessment of Functioning score in CHR (r = .35, P < .05). This observed pattern of white matter connectivity disruptions in FEP and in CHR suggests that this pattern of disconnectivity not only highlights the involvement of thalamus but also might be useful as an early biomarker for psychosis.