Emotion dysfunction has long been considered a cardinal feature across psychotic disorders, including schizophrenia and affective psychosis. However, few studies have used objective markers of emotional function to compare psychotic disorders to one another, and fewer studies have examined such markers within a longitudinal framework. Here, we examine one objective marker of emotional responsivity, the late positive potential (LPP), which is a centro-parietal event-related potential (ERP) that tracks the dynamic allocation of attention to emotional vs. neutral stimuli. We used the LPP to characterize abnormal emotional responsivity by relating it to negative, depressive, and psychotic symptoms among two clinical groups: individuals diagnosed with affective psychosis and individuals with schizophrenia. We also used a long-term longitudinal framework, examining concurrent associations between LPP amplitude and symptom severity, as well as prospective associations with symptoms 4 years later. Participants were 74 individuals with psychotic illness: 37 with schizophrenia spectrum disorders and 37 with a primary affective disorder (psychotic bipolar disorder, psychotic depression). There were no mean-level differences in LPP amplitude between the schizophrenia spectrum and primary affective psychosis group. In the primary affective psychosis group, reduced LPP amplitude was associated with greater depressive, negative, and psychotic symptom severity, both concurrently and at follow-up; associations between LPP and symptoms were not observed within the schizophrenia spectrum group. This pattern of results suggests that the neural correlates of emotion dysfunction may differ across psychotic disorders. One possibility is that schizophrenia is characterized by a decoupling of symptom severity and emotional processing. Such findings underscore the importance of analyzing transdiagnostic samples to determine common or specific symptom relationships across various patient populations.