Background: Lumbar puncture is among the investigations used to determine the aetiology and prognosis of cardiac arrest, despite a dearth of data on its performance. We aimed to assess the diagnostic and prognostic performance of lumbar puncture after cardiac arrest. Methods: We retrospectively studied data from prospectively established databases of consecutive patients who were admitted to two French ICUs in 2007-2016 with sustained return of spontaneous circulation (ROSC) after cardiac arrest and who underwent lumbar puncture as an aetiological investigation. Results: Of 1984 patients with sustained ROSC, 65 (3.3%) underwent lumbar puncture and were included. Lumbar puncture identified a neurological cause of cardiac arrest in 6/65 (9%) patients, including 5 with neurologic prodromal symptoms before cardiac arrest and 4 with the lumbar puncture done post-mortem. Lumbar puncture performed before death showed nonspecific cerebrospinal fluid abnormalities in 37/53 (69.8%) patients. By univariate analysis, factors significantly associated with cerebrospinal fluid abnormalities were shorter no-flow time (0 min [0-3] versus 4 min [2-10], p=0.004) and post-resuscitation shock (26 [70%] versus 5 [31%], p<0.01). Presence of cerebrospinal fluid abnormalities was non significantly associated with poorer outcomes (CPC 3-4-5) (p=0.06). Conclusions: Lumbar puncture, although rarely performed, can contribute to the aetiological diagnosis of cardiac arrest. As a second-line investigation, it identified the cause in 9% of our patients. Nonspecific cerebrospinal fluid abnormalities are common after cardiac arrest, perhaps due to blood-brain barrier disruption, and may carry prognostic significance.