Current classification systems of mental illness characterize bipolar disorder (BD) and schizophrenia (SCZ) as discrete diagnostic categories, presupposing distinct presentation, etiology, and treatment, despite mounting evidence of epidemiological and genetic overlap between the two and in a way that does not fully reflect increasing knowledge of underlying biological mechanisms or pathogenesis. To further elucidate the nature of phenotypic overlap versus differentiation between the two, emotion perception and regulation deficits were examined via a task of affect recognition during fMRI. Data were collected at the Karolinska Institute in Sweden; demographically matched participants were recruited based on national medical records data in line with study protocol approved by Karolinska and UCLA IRB. Behaviorally, patients with SCZ (n=41) and BD (n=38) demonstrated similar impairment in affect labeling relative to controls (n=64); however, SCZ patients showed greater deficits during affect matching and the two groups showed differences in corresponding patterns of neural activation. During affect matching, whole-brain voxel-wise BOLD signal analysis indicated both patient groups showed hypoactivation relative to controls in putative social cognitive network regions but the specific regions differed by group, such that BD patients showed hypoactivation of posterior cingulate/precuneus, whereas SCZ patients showed hypoactivity in right amygdala/hippocampus. In addition, the SCZ group demonstrated failure of fronto-limbic circuitry to modulate ventral face and emotion processing regions during affect labeling; they showed hyperactivation of fusiform gyrus, inferior occipital cortex, and posterior superior and middle temporal gyrus and did not show negative functional connectivity between these regions as shown in controls and BD patients through PPI analysis. SCZ patients also showed aberrant positive cortico-cortical connectivity in frontal regions versus BD patients, suggestive of compensatory recruitment of additional frontal regions. The current study thus adds new and novel evidence to the ongoing debate regarding the utility of categorical classification of disease, demonstrating underlying disparateness in neurophysiology related to specific aspects of the socio-emotional domain and lending at least partial validation to the current diagnostic distinction. Implications for treatment considerations are also discussed.