For the case of pharyngeal carcinomas, the clinical value as well as the stability of several evaluation methods of MR tomographic perfusion measurement are compared. Eighteen patients suffering from histologically proven squamous cell carcinomas were investigated by MR tomography (1.5 T, 0.2 mmol/kg Gd-DTPA) prior to and during radiation therapy. Perfusion measurements were performed using a double-echo FLASH sequence. Parameters describing regional blood flow, blood volume, mean transit time, and interstitial concentration of contrast medium (CM) were calculated, applying seven different combinations of correction approaches (separating the shortening of T1 and T2*, arterial input function (AIF), and tumor shunts). Their correlations to MR independent tumor physiological parameters were analyzed (metabolic activity measurements using 18F-FDG-PET, polarographical pO2 measurement, tumor volume). Significant improvements of the correlation between perfusion-dependent and other tumor physiological parameters could be achieved by decoupling the shortening of T1 and T2* and by applying of the tumor shunt model. Deconvolution from the AIF deteriorated the correlation. Therefore, the elimination of the T1 shortening due to interstitial CM proves to be essential for MR perfusion measurements on contrast medium uptaking lesions. Depending on the measurement conditions (temporal resolution, signal-to-noise ratio), the consideration of the AIF can even make the results significantly worse by introducing additional measuring errors.