Purpose: Cerebrovascular reactivity (CVR) is impaired in adolescents with cardiovascular disease risk factors. A breath-hold test is a noninvasive method of assessing CVR, yet there are no reliability data of this outcome in youth. This study aimed to assess the reliability of a breath-hold protocol to measure CVR in adolescents. Methods: Twenty-one 13 to 15 year old adolescents visited the laboratory on two separate occasions, to assess the within-test, within-day and between-day reliability of a breath-hold protocol, consisting of three breath-hold attempts. CVR was defined as the relative increase from baseline in middle cerebral artery mean blood velocity following a maximal breath-hold of up to 30 seconds, quantified via transcranial Doppler ultrasonography. Results: Mean breath-hold duration and CVR were never significantly correlated (r < .31, P > .08). The within-test coefficient of variation for CVR was 15.2%, with no significant differences across breath-holds (P = .88), so the three breath-hold attempts were averaged for subsequent analyses. The within-and between-day coefficients of variation for CVR were 10.8% and 15.3%, respectively. Conclusions: CVR assessed via a three breath-hold protocol can be reliably measured in adolescents, yielding similar within-and between-day reliability. Analyses revealed that breath-hold length and CVR were unrelated, indicating the commonly reported normalization of CVR to breath-hold duration (breath-hold index) may be unnecessary in youth. K E Y W O R D S cerebral blood flow, endothelial function, hypercapnic stimulus, reproducibility, transcranial Doppler ultrasound, youth 1 | INTRODUCTION Cerebrovascular reactivity (CVR) refers to the ability of the human brain to modulate cerebral blood flow in response to changes in stimuli, such as the partial pressure of arterial carbon dioxide (PaCO 2). Impairments in CVR are an important hallmark for cerebrovascular disease (CVD) progression. Research highlights that impairments in CVR in adults is associated with Alzheimer's disease, 1 neurocognitive decline, 2 stroke, 3,4 and independently predicts future CVD events in patients with CVD risk factors. 5 Impairments in CVR are present in