This manuscript describes a study of the evaluation of T2 and T2* changes in a hypervascular tumor model (hemodialyzer phantom) under conditions simulating dynamic perfusion study. The measured 1/T2* was strongly dependent on the compartmentalization of contrast material within the model, whereas the observed 1/T2 was not. Gradient-echo magnetic resonance (MR) imaging may be more suitable than a spin-echo MR imaging for the evaluation of tumor vascularity.