BOLD fMRI is becoming the preferred modality to study cerebrovascular reactivity. Amongst the vasoactive stimuli to induce it, breath holding is considered equal to other methods but both cerebrovascular reserve measures and breath holding designs lack standardization. We have systematically considered whether breath holding paradigms at short time periods are both a practical as well as equally effective and robust alternative to the elaborate CO2-inhalation in estimating cerebrovascular reactivity. Using BOLD fMRI, eleven healthy subjects were examined with dynamic T2*-weighted echo planar imaging at 3 T. Cerebrovascular reactivity was measured during episodes of breath holding for 10 s and 20 s, both after deep inspiration and after extended expiration, and compared to the reactivity following inhalation of an air mixture with 5% CO2-content. 20 s breath holding challenges yielded higher reactivities represented by corresponding higher BOLD-signal changes when compared to 10 s ones, as did expiratory paradigms when compared to inspiratory ones. Mean signal changes ranged from 2% (10 s breath holding after inspiration) to 3% (20 s breath holding after expiration). The CO2-challenge yielded an about threefold higher reactivity (6%). The breath holding tasks evoked BOLD-signal increases which compare favorably to published data. A 20 s breath holding period appears to be ideal and is tolerated well. This experimental design is thus practical, effective and robust. To further standardize the breath holding design and the measurement of cerebrovascular reserve, controlled breathing during resting periods as well as modelling of the fMRI-signal to end-tidal CO2-levels should be employed in future studies.