Objective: To conduct a meta-analysis of Theory of Mind studies exclusively in euthymic patients with bipolar disorder. Method: After the exclusion of studies evaluating symptomatic patients during acute episodes, we performed a meta-analysis including a total of 30 studies, comparing 1294 euthymic bipolar disorder patients and 1116 healthy controls. Results: Patients with bipolar disorder presented a significant impairment in Theory of Mind performance when compared to controls (Hedge’s g = −0.589, 95% confidence interval: −0.764 to −0.414, Z = −6.594, p < 0.001). When compared to controls, Theory of Mind was impaired in patients with both bipolar disorder I (Hedge’s g = −0.663, 95% confidence interval: −0.954 to −0.372, Z = −4.462, p < 0.001) and bipolar disorder II (Hedge’s g = −1.165, 95% confidence interval: −1.915 to −0.415, Z = −3.044, p = 0.002). Theory of Mind impairments were also significantly more severe in verbal tasks (Hedge’s g = −1.077, 95% confidence interval: −1.610 to −0.544, Z = −3.961 p < 0.001) than visual tasks (Hedge’s g =−0.614, 95% confidence interval: −0.844 to −0.384, Z = −5.231, p < 0.001) when compared to controls. Conclusion: The results obtained confirm that Theory of Mind is impaired in remitted bipolar disorder patients, being a potential endophenotype for bipolar disorder. Moreover, we found higher deficits in verbal Theory of Mind, compared with visual Theory of Mind. Since most studies were cross-sectional, there is a need for longitudinal studies to evaluate whether the deficits detected in Theory of Mind are progressive over the course of the illness.