Multi-parametric quantitative BOLD (mq-BOLD) measurements have been successfully applied in several studies to assess the vascular oxygenation, however, calculated relative Oxygen Extraction Fraction (rOEF) shows unphysiological systematic elevations. We suspect biased T2-measurements to be the main source of error. Therefore, we present an optimized 3D-GraSE T2-mapping sequence and its evaluation in four stages within phantoms, young healthy controls, elderly controls and internal carotid artery stenosis (ICAS) patients. We found significant T2-decreases, fully consistent with reference-values, and thereby significantly decreased rOEF-values by ~25 % towards physiologically more realistic values. Additional clinical value was demonstrated by detecting focal rOEF-increases in an ICAS-patient.