Objective: Bleeding after cardiac surgery leads to poor outcomes. The objective of the study was to build the PeriOperative Bleeding Score in Cardiac surgery (POBS-Card) to predict bleeding after cardiac surgery. Methods: We conducted a retrospective cohort study in 2 academic hospitals (2016-2019). Inclusion criteria were adult patients after cardiac surgery under cardiopulmonary bypass. Exclusion criteria were heart transplantation, assistance, aortic dissection, and preoperative hemostasis diseases. Bleeding was defined by the universal definition for perioperative bleeding score ≥2. POBS-Card score was built using multivariate regression (derivation cohort, one center). The performance diagnosis was assessed using the area under the curve in a validation cohort (2 centers) and compared with other scores. Results: In total, 1704 patients were included in the derivation cohort, 344 (20%) with bleeding. Preoperative factors were body mass index <25 kg/m2 (odds ratio [OR], 1.48 [1.14-1.93]), type of surgery (redo: OR, 1.76 [1.07-2.82]; combined: OR, 1.81 [1.19-2.74]; ascendant aorta: OR, 1.56 [1.02-2.38]), ongoing antiplatelet therapy (single: OR, 1.50 [1.09-2.05]; double: OR, 2.00 [1.15-3.37]), activated thromboplastin time ratio >1.2 (OR, 1.44 [1.03-1.99]), prothrombin ratio <60% (OR, 1.91 [1.21-2.97]), platelet count <150 g/L (OR, 1.74 [1.17-2.57]), and fibrinogen <3 g/L (OR, 1.33 [1.02-1.73]). In the validation cohort of 597 patients, the area under the curve was 0.645 [0.605-0.683] and was superior to other scores (WILL-BLEED, Papworth, TRUST, TRACK). A threshold >14 predicted bleeding with a sensitivity of 50% and a specificity of 73%. Conclusions: POBS-Card score was superior to other scores in predicting severe bleeding after cardiac surgery. Performances remained modest, questioning the place of these scores in the perioperative strategy of bleeding-sparing.