BackgroundScarce research has focused on Visual Memory (VM) deficits as a possible neurocognitive endophenotype of bipolar disorder (BD). The main aim of this longitudinal, family study with healthy controls was to explore whether VM dysfunction represents a neurocognitive endophenotype of BD. MethodsAssessment of VM by Rey–Osterrieth Complex Figure Test (ROCF) was carried out on a sample of 317 subjects, including 140 patients with BD, 60 unaffected first-degree relatives (BD-Rel), and 117 genetically-unrelated healthy controls (HC), on three occasions over a 5-year period (T1, T2, and T3). BD-Rel group scores were analyzed only at T1 and T2. ResultsPerformance of BD patients was significantly worse than the HC group (p < 0.01). Performance of BD-Rel was also significantly different from HC scores at T1 (p < 0.01) and T2 (p = 0.05), and showed an intermediate profile between the BD and HC groups. Only among BD patients, there were significant differences according to sex, with females performing worse than males (p = 0.03). Regarding other variables, education represented significant differences only in average scores of BD-Rel group (p = 0.01). LimitationsImportant attrition in BD-Rel group over time was detected, which precluded analysis at T3. ConclusionsBD patients show significant deficits in VM that remain stable over time, even after controlling sociodemographic and clinical variables. Unaffected relatives also show stable deficits in VM. Accordingly, the deficit in VM could be considered a potential endophenotype of BD, which in turn may be useful as a predictor of the evolution of the disease. Further studies are needed to confirm these findings.