Patients with unstable pelvic fractures combined with traumatic brain injury (TBI) present complex conditions, with pathophysiological contradictions in clinical treatment between limited fluid resuscitation and maintenance of cerebral perfusion pressure. A retrospective analysis was conducted on 204 patients with unstable pelvic fractures combined with TBI between January 1, 2010, and May 31, 2025. A Nomogram model was constructed via LASSO and multivariate logistic regression, with simultaneous development of an interpretable machine learning prediction model. Results identified shock, abnormal systolic blood pressure (SBP), non-surgical management, and prolonged coagulation time as independent poor prognosis risk factors. The Nomogram constructed using SBP, shock, pelvic fracture surgery, and prothrombin time (PT), showed good performance (AUC = 0.801, C-index = 0.7747). Poor prognosis risk was lowest at SBP 110-122 mmHg; a U-shaped correlation existed between SBP and prognosis. The Nomogram and interpretable XGBoost model provide a basis for individualized blood pressure management, coagulation monitoring, and surgical decision-making. ChiCTR2500103926, registered on May 9, 2025, PID: 275507.