PURPOSE: To investigate variation in dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) pharmacokinetic parameter measurements between different methods of precontrast tissue relaxation (T10) estimation: pixel-based mapping versus a fixed reference value. MATERIALS AND METHODS: In 15 DCE-MRI studies the female pelvis, uterine fibroids, the left psoas muscle, and the fifth lumbar vertebral body were chosen to represent tissues with varying perfusion characteristics. All DCEMRI studies were processed using a variable flip angle T10 map and a fixed T10 reference value of 1000 msec. A subset of five DCE-MRI studies were each processed multiple times using the fixed T10 method with the reference T10 ranging from 0–2000 msec in 100-msec increments. Pharmacokinetic measurements of Ktrans, kep, ve, and initial area under the gadolinium curve (iAUGC) were performed maintaining the identical position for region of interest placement on each structure. RESULTS: The mean difference in pharmacokinetic output between the pixel-based T10 map and the fixed T10 reference value ranged from 6.6% for kep in the muscle to 54.9% for iAUGC in the vertebral body. At lower T10 (<1000 msec) aberrations in T10 estimation resulted in a larger error. Accurate measurement of T10 for each structure subsequently incorporated as a fixed T10 reference value yielded relative differences from 41.8% to 22.3% compared to the pixel-based T10 map. CONCLUSION: Direct comparison of pharmacokinetic parameters derived from a pixel-based approach versus a reference value uniformly applied to all pixels for T10 estimation is impeded by the inherent spatial heterogeneity of T10 within tissues.