Background: Attention deficits are at the core of both schizophrenia and bipolar disorders. However, few studies have compared both patient groups with each other and healthy controls simultaneously. The current study aims to compare Continuous Performance Test - Identical Pairs (CPT-IP) performance between these groups on measures of d’, hits, false alarms, random responses and reaction time. The groups are also compared on a more novel measure of intra-individual variability, i.e. reaction time Coefficient of Variation (RT CoV). Additionally, the association between symptom load and CPT performance in patient groups will be investigated. Methods: One hundred and forty-two patients with schizophrenia, 68 patients with bipolar disorder I, and 637 healthy controls completed a CPT-IP as part of the cognitive assessment at the Thematically Organized Psychosis (TOP) study at the Norwegian Centre for Mental Disorders Research (NORMENT). Global functioning, IQ, and symptom load were also assessed. Data were analyzed using repeated measures ANOVA and Pearson correlations. Results: Healthy controls performed significantly better than the patient groups on all CPT-IP measures, except false alarms and RT CoV. The patient groups only differed significantly on the number of hits, where the bipolar disorder group outperformed the schizophrenia group. Symptom load was not significantly associated with CPT performance. Conclusion: The findings support the assumption of sustained attention deficits in both schizophrenia and bipolar disorder, but the lack of patient group differences is not in line with earlier research. Inconclusive RT CoV results highlight the need for additional studies employing RT CoV as a measure of information-processing instability. The lack of associations between symptom load and CPT performance could indicate a shared, underlying factor affecting CPT performance in both patient groups.