To investigate the prevalence and behavior of sleep-related breathing disorders (SRBDs) associated with a first-ever stroke or transient ischemic attack (TIA), we prospectively studied 161 consecutive patients admitted to our stroke unit. Complete neurological assessment was performed to determine parenchymatous and vascular localization of the neurological lesion. Stroke subtype was categorized as TIA, ischemic (IS), or hemorrhagic (HS). A portable respiratory recording (PRR) study was performed within 48-72 h after admission (acute phase), and subsequently after 3 mo (stable phase). During the acute phase, 116 patients (71.4%) had an apnea-hypopnea index (AHI) > 10 events/h and 45 (28%) had an AHI > 30. No relationships were found between sleep-related respiratory events and the topographical parenchymatous location of the neurological lesion or vascular involvement. Cheyne-Stokes breathing (CSB) was observed in 42 cases (26.1%). There were no significant differences in SRBD according to the stroke subtype except for the central apnea index (CAI). During the stable phase a second PRR was performed in 86 patients: 53 of 86 had an AHI > 10 and 17 of 86 had an AHI > 30. The AHI and CAI were significantly lower than those in the acute phase (16.9 +/- 13.8 versus 22.4 +/- 17.3 and 3.3 +/- 7.6 versus 6.2 +/- 10.2, respectively) (p < 0.05) while the obstructive apnea index (OAI) remained unchanged. CSB was observed in 6 of 86 patients. The prevalence of SRBD in patients with first-ever stroke or TIA is higher than expected from the available epidemiological data in our country. No correlation was found between neurological location and the presence or type of SRBD. Obstructive events seem to be a condition prior to the neurological disease whereas central events and CSB could be its consequence.