This dissertation has developed, implemented and tested a novel computer based system (AUTOALIGN) that incorporates an algorithm for the alignment of functional Magnetic Resonance Image (fMRI) time series. The algorithm assumes the human brain to be a rigid body and computes a head coordinate system on the basis of three reference points that lie on the directions correspondent to two of the eigenvectors of inertia of the volume, at the intersections with the head boundary. The eigenvectors are found weighting the inertia components with the voxel's intensity values assumed as mass. The three reference points are found in the same position, relative to the origin of the head coordinate system, in both test and reference brain images. Intensity correction is performed at sub-voxel accuracy by tri-linear interpolation. A test fMR brain volume in which controlled simulations of rigid-body transformations have been introduced has preliminarily assessed system performance. Further experimentation has been conducted with real fMRI time series. Rigid-body transformations have been retrieved automatically and the values of the motion parameters compared to those obtained by the Statistical Parametric Mapping (SPM99), and the Automatic Image Registration (AIR 3.08). Results indicated that AUTOALIGN offers subvoxel accuracy in correcting both misalignment and intensity among time points in fMR images time series, and also that its performance is comparable to that of SPM99 and AIR3.08.