A two-dimensional T2*-weighted gradient-echo sequence was used to image the rat brain before and during graded hypoxemia. Changes in R2* (deltaR2*) with respect to the control state were calculated for brain parenchyma and were compared with changes in hemoglobin saturation measured from both arterial and jugular venous blood samples. DeltaR2* was first correlated with the changes in arterial (deltaYa) and venous (deltaYv) hemoglobin saturations individually. Although a general trend toward a linear relationship with deltaR2* was observed for both deltaYa and deltaYv, neither alone was strong (correlation coefficients r=0.71 and 0.75 for deltaYa and deltaYv, respectively, and standard errors of the regression (SER)=0.52 and 0.48 for deltaYa and deltaYv, respectively). However, when an "effective" cerebral blood hemoglobin saturation change (deltaYb) was constructed that takes into account the approximate weighting of the contributions from the arterial and venous phases of the circulation (deltaYb = 0.75 x deltaYv + 0.25 x deltaYa), a stronger correlation with deltaR2* was obtained and there was less variance (r=0.87 and SER=0.35). It is concluded that an appropriate weighting of the contributions of arterial and venous phases of the circulation must be taken into account in modeling the volume susceptibility effects of deoxyhemoglobin on R2* of brain parenchyma. In this way, a more accurate relationship between deltaR2* and deltaYb can be obtained.