A 76-year-old woman with unspecified congenital heart disease was admitted on April 25th for TIA. She had a possible history of atrial fibrillation. A slight fever was noted on admission. Her ECG was abnormal, as well as her transthoracic echocardiography (TTE). Troponin I was slightly increased. On May 11th, a stroke occurred, in relation with an occlusion of the basilar artery. The patient was transferred to our institution for an emergency desobstruction. A dramatic improvement allowed her to be discharged to a rehabilitation center on May 18th. However, she was re-hospitalized on June 5th, due to sepsis and neurological worsening. MRI showed new ischemic brain lesions. Several episodes of paroxysmal atrial fibrillation were documented, as well as pulmonary hypertension. Effective heparin therapy was initiated and transesophageal echocardiography (TEE) was requested this time. It revealed a congenital valvular heart disease (a subaortic membrane), complicated by infective endocarditis. Despite a monitoring of aPTT, a fatal hemorrhagic shock occurred. We report this unfortunately remarkable case to address the following important points: 1) In the setting of a neurological event, abnormal ECG and/or abnormal TTE and/or Troponin I elevation may indicate a cardioembolic mechanism and therefore seek a cardiac source of embolism. 2) When TTE fails to identify a cardiac source of embolism, TEE should be performed, especially when a preexisting heart disease is suspected or known. 3) The multiplicity in space (infarcts in both the anterior and posterior circulation, or bilateral) and/or the multiplicity in time (infarcts of different age) may indicate a cardioembolic stroke. 4) Congenital subaortic membrane predisposes to infective endocarditis. 5) When anticoagulant therapy is initiated on strong arguments in a septic patient (much discussed in infective endocarditis), aPTT monitoring alone may not be enough. An anti-Xa monitoring may be more appropriate.