The European Stroke Organization recommends performing lumbar puncture (LP) in the etiological investigation of an ischemic stroke ‘‘when indicated’’ [1]. This recommendation leaves the decision to the clinician. Cerebrospinal fluid analysis may yield essential information in cases of pediatric ischemic stroke [2]. In adults it is generally accepted that a LP should be performed in case of suspicion of central nervous system (CNS) infection and vasculitis [1, 3], but its role in the etiological investigation of an ischemic stroke is unclear. We aimed to determine the frequency of LP performed in the investigation of ischemic stroke of young patients, the reasons that lead to its performance, CSF results and their contribution to the etiological diagnosis. We analyzed a retrospective cohort of consecutive patients with ischemic stroke, aged 18–55 years, admitted to Santa Maria Hospital Stroke Unit from May 2007 to December 2009. Patients were investigated according to a standardized protocol that includes brain CT and/or MRI, electrocardiogram, laboratory evaluation with vasculitis and infection screening (HIV, hepatitis, VDRL), cervical arteries and transcranial ultrasound, transthoracic/transesophagic echocardiogram and 24-h ECG Holter monitoring. The decision to perform digital subtraction angiography and LP was individually discussed. CSF analysis always included: gross characteristics, hemoglobin, total cell count, total protein, glucose, chloride and immunoglobulins; specific antibody indices were integrated when appropriate. Clinical files from patients who were submitted to a LP were retrospectively reviewed. Demographic, clinical, analytical and imagiological features and stroke etiology according to the TOAST classification were analyzed. From the 143 young stroke patients admitted during the study period, a lumbar puncture was performed in 36 (25.2%). Four patients were excluded because the LP was done for the differential diagnosis (stroke vs. other condition) at the emergency department. Reasons that led to LP were: (1) suspicion of infection (one HIV?, two VDRL? and one endocarditis), (2) suspicion of CNS vasculitis (four ANAs?, one antiphospholipid?, six with intracranial arterial stenosis), and (3) cryptogenic stroke (nine) or with a PFO as the only identified cause (eight). Twenty-nine out of 32 analyzed patients had completely normal CSF. Three patients showed [5 cells/ml, three had CSF proteins [50 mg/dl, and matched CSF oligoclonal bands were present in one patient; CSF VDRL was positive in one patient. Post LP headache occurred in nine patients, but there were no other complications related to the procedure. LP contributed to a change in the etiological ischemic stroke classification in one patient (3%, 95% CI 1–16) from undetermined cause to other determined cause (syphilis). Variation in the proportion of etiological subgroups found in young adult stroke studies reflects different ancillary investigations [4]. Regarding LP in ischemic stroke etiological investigations, apart from studies before the era of brain or cardiac imaging [5–7], there is little information in the literature. However, though inflammatory causes of stroke such as Lyme neuroborreliosis [8] among others are rare, correct diagnosis and treatment of these rare causes R. Geraldes A. C. Fonseca P. Canhao T. P. Melo J. M. Ferro Stroke Unit, Department of Neurosciences, Santa Maria Hospital, Av. Prof. Egas Moniz, 1649-035 Lisbon, Portugal