This paper introduces a novel approach to multimodality medical image registration using Mutual Information (MI) as a matching criterion. The method maximizes MI to measure the statistical dependence between corresponding voxel intensities in two images, assuming that geometric alignment maximizes this dependence. The accuracy and robustness of the MI criterion are evaluated for rigid body registration of CT, MR, and PET images. The results demonstrate subvoxel accuracy and robustness to various implementation issues and image content, making the method suitable for clinical applications. The paper also discusses the theoretical foundation of MI, the algorithm implementation, and the evaluation of accuracy and robustness through experiments with different datasets. The method's advantages include its data independence, automatic nature, and minimal user interaction, making it well-suited for clinical practice.This paper introduces a novel approach to multimodality medical image registration using Mutual Information (MI) as a matching criterion. The method maximizes MI to measure the statistical dependence between corresponding voxel intensities in two images, assuming that geometric alignment maximizes this dependence. The accuracy and robustness of the MI criterion are evaluated for rigid body registration of CT, MR, and PET images. The results demonstrate subvoxel accuracy and robustness to various implementation issues and image content, making the method suitable for clinical applications. The paper also discusses the theoretical foundation of MI, the algorithm implementation, and the evaluation of accuracy and robustness through experiments with different datasets. The method's advantages include its data independence, automatic nature, and minimal user interaction, making it well-suited for clinical practice.