The BOLD response to a gas with fixed inspired CO2 concentration is commonly used to assess the health of the cerebral vasculature. Cerebrovascular reactivity (CVR) magnitude repeatability has been assessed in previous studies, but without reporting potential physiological confounders such as heart and respiration rates. Furthermore, less is known about CVR delay repeatability. In this work, we assessed the within-day test-retest repeatability of both quantities in healthy volunteers at 3 T while measuring the heart and respiration rates. We found that both show good repeatability, though CVR delay estimates are unreliable in tissues with low contrast to noise ratio.