Spatio-temporal reconstruction methods for positron emission tomography estimate the tissue time-activity curves that are required for functional imaging. Since the time-related change of activity at a voxel has a temporal correlation by itself, temporal basis function approaches can be adopted. However, these image reconstruction methods suffer from a high computational cost. We have proposed a novel spatio-temporal reconstruction method using a temporal basis function approach, which is based on a fast block-iterative algorithm named Dynamic Row-Action Maximum-Likelihood Algorithm (DRAMA). Using the proposed method, data quickly converge to an estimate after around two iterations. This study aimed to validate the performance of the proposed algorithm on small animal PET data. We applied the proposed method to ¹⁸F-FDG mouse dynamic PET data. A parametric image of regional glucose metabolism reconstructed from the proposed method was almost identical to those obtained from conventional reconstruction algorithms. The proposed method took 14.3 h for computation, which was twice as fast as conventional algorithms. These results support the usability of the proposed algorithm for voxel-by-voxel estimation.