We compared standard Tofts model with a two-tissue compartment model (2TCM) of dynamic contrast enhanced (DCE) MRI for diagnosis of prostate cancer. The 2TCM has one slow and one fast exchanging compartment. The standard Tofts model parameters (Ktrans and kep) were compared with the 2TCM parameters (Kitrans and kiep, i=1,2). There was a strong correlation between Ktrans and K1trans for cancer, but weak correlation between kep and k1ep. This demonstrated that the Tofts model often does not fit contrast agent concentration curves accurately, and the 2TCM can provide new diagnostic information with fewer false positives in diagnosis of prostate cancer.