Combined with medical robotic system, mechanical property characterization of the prostate could enhance the diagnosis and localization of prostate cancer. However, despite the importance of localization, complex geometry and boundary conditions of the prostate make localization difficult. This paper proposes a method for localizing abnormality using finite element modeling of the prostate glands. Ex-vivo experiment was performed to resected human prostate tissue for six regions of the prostate using robotic system which was previously designed to induce sweeping palpation. For each region, three dimensional prostate computational models from the CT image segmentation were developed for mechanical property characterization and local property criteria construction. The measured force responses were applied to the prostate model fittings and mechanical property of ex-vivo experiments were estimated. By using comparative study with local property criteria from the prior indentation study, suspicious regions were localized. The results showed that the suspicious regions were discriminated and the sensitivity was 66.7%.