Recently a new time-domain method has been presented for deconvolution . This multidimensional method completely eliminates the problems of the previous methods in one dimension and covers a reasonable part of the solutions in multidimensions. In this paper, we present some of the properties of this method. We will especially focus on the frequency domain behaviour of the algorithm as well as the performance under numerical errors and errors due to noise. In addition we will present examples of the applications including blind deconvolution with a modified NAS-RIF algorithm.