A new state estimation scheme is presented for multidimensional dynamic systems with arbitrary independent interference and noises, and is based upon quantization, multiple composite hypothesis testing, and a suboptimum decoding technique of Information Theory. The estimation of the state vector is sequentially done, component-by-component, in parallel and in blocks. “Component-by-component” estimation results in a considerable memory reduction for the implementation of the scheme, while estimation is blocks makes the implementation independent of time. Simulation results, some of which are presented, have shown that the new scheme performs well, whereas the classical estimation techniques are not, in general, applicable to the state estimation problem in the presence of arbitrary interference.