Patients with schizophrenia have difficulty in decoding facial affect. A study using event-related functional neuroimaging indicated that errors in fear detection in schizophrenia are associated with paradoxically higher activation in the amygdala and an associated network implicated in threat detection. Furthermore, this exaggerated activation to fearful faces correlated with severity of flat affect. These findings suggest that abnormal threat detection processing may reflect disruptions between nodes that comprise the affective appraisal circuit. Here we examined connectivity within this network by determining the pattern of intercorrelations among brain regions (regions of interest) significantly activated during fear identification in both healthy controls and patients using a novel procedure CORANOVA. This analysis tests differences in the interregional correlation strength between schizophrenia and healthy controls. Healthy subjects' task activation was principally characterized by robust correlations between medial structures like thalamus (THA) and amygdala (AMY) and middle frontal (MF), inferior frontal (IF), and prefrontal cortical (PFC) regions. In contrast, schizophrenia patients displayed no significant correlations between the medial regions and either MF or IF. Further, patients had significantly higher correlations between occipital lingual gyrus and superior temporal gyrus than healthy subjects. These between-group connectivity differences suggest that schizophrenia threat detection impairment may stem from abnormal stimulus integration. Such abnormal integration may disrupt the evaluation of threat within fronto-cortical regions.