We studied the effect of mean arterial blood pressure (mBP) on the cerebrovascular CO2 reactivity (K) by measuring the middle cerebral artery blood flow velocity (Vm) with transcranial Doppler ultrasonography (TCD) in 24 anesthetized patients. By measuring Vm during hypocarbia (po) and hypercarbia (er) at mBP of 60-120 mmHg, three kinds of K value were evaluated as follows: K(h), K(l), or K(c) = delta lnVm/delta PaCO2 = [lnVm(er)-lnVm (po)]/[PaCO2(er)-PaCO2 (po)], where K(h) was calculated between [Vm(er), PaCO2(er)] at a high mBP and [Vm(po), PaCO2(po)] at a low mBP; K(l), between [Vm(er), PaCO2(er)] at a low mBP and [Vm(po), PaCO2(po)] at a high mBP; K(c), at the same mBP corrected. The relationship between Vm and mBP showed a linear regression at each constant level of PaCO2. The K(h) shifted higher and the K(l) shifted lower in comparison with the K(c) following the widening of the mBP difference. The K(l) was affected especially by small variations of mBP. In conclusion, to evaluate simply the cerebrovascular CO2 reactivity modulated by mBP with TCD, K(h) may be acceptable as long as it is realized that the K(h) tends to be relatively high K value.