Cerebrovascular reactivity (CVR) of middle cerebral artery velocity (MCAv) to CO 2 is a common method to assess cerebrovascular function. Yet, the approaches used to calculate CVR outcomes vary. The aim of this study was to explore the withinand between-day reliability of traditional CVR outcomes. The second aim was to explore the reliability of novel kinetic-based analyses. Healthy adults (n = 10, 22.3 ± 3.4 years) completed assessments of CVR over 4 min using a fixed fraction of inspired CO 2 (6%). This was repeated across four separate visits (between-day), and on one visit measures were repeated 2.5 h later (within-day). No mean biases were present between assessments for traditional CVR metrics, expressed as absolute (cm/s/mmHg) or relative (%/mmHg) outcomes (minute 3, minute 4, peak 1 s, peak 30 s) (betweenday: P > 0.14, η p 2 < 0.20; within-day: P > 0.22, d > 0.27). Absolute, rather than relative, CVR yielded the most reproducible parameters (coefficient of variation: 8.1-13.2%vs. 14-83%, respectively). There were significant differences between CVR outcomes (P < 0.001, η p 2 > 0.89) dependent on the time point used to determine CVR, as a steady state MCAv response was rarely observed. Furthermore, the MCAv response was not reproducible within an individual (κ = 0.15, P = 0.09). No mean differences were present for novel kinetic outcomes (amplitude, time-delay, time constant) (between-day: P > 0.05, d < 0.33; within-day: P > 0.38, d < 0.25). The results support the need for standardisation and indicate CVR should be defined as a dynamic peak, rather than a set time point for increased reliability. For novel kinetic outcomes variability was greater (CV: 8.7-120.9%) due to the nature of time-based metrics.