Bolus tracking magnetic resonance imaging (MRI) is a powerful technique for assessing cerebral perfusion, but its capability to measure absolute cerebral blood flow (CBF) and volume (CBV) values is still debated. To validate the MRI technique, absolute CBF and CBV values in healthy humans obtained by echo planar gradient echo MRI were compared to H(2)(15)O and (11)CO positron emission tomography (PET) before and after acetazolamide (ACZ) (n = 8) or saline (n = 4) administration. The repeatability of CBF and CBV measurements was moderate with both methods, and slightly lower with MRI than with PET. At rest, the mean CBF values were similar with both techniques except in the cortex where they were moderately higher with MRI. CBV was higher with MRI than with PET in all areas, which may reflect an underestimation of the arterial input function (AIF). After ACZ, a significant CBF increase was observed in gray matter with both MRI and PET, suggesting that MRI might be used to assess the cerebrovascular reserve. In individual subjects, the correlation between MRI and PET measurements was good for both CBF and CBV (R(2) between 0.70 and 0.84). However, when all results were considered as a group, R(2) was lower (0.40 to 0.65), and the limits of agreement between the two methods (SD of the difference) were large. Our data suggest that physiologic CBF values and systematically overestimated CBV values may be obtained with MRI in healthy humans, but that an individual scale factor should be applied to MRI measurements to improve the agreement with PET.