An interesting implementation procedure is developed for use in a covariance propagation problem involving a discrete Kalman filter. Using a matrix continued fraction (!\olCF) method for the calculation of the covariance matrix, the final irnplementation requires one matrix inversion involving a sequence of controllability matrices, observability matrices, and an initial condition matrix. Using procedures as discussed by Mendel (1971), Bierman (1973, 1977(1971) and
Carlson (1973), a numerical comparison in terms of computing timo and computer storage is made between the approach developed here and the conventional method.This algorithm is intended to be used as an alternative method to implement the Kalman filter. For certain classes of large-scale systems, the MCF method provides both a reduction of computer time and computer storage. An example is worked to illustrate the applicability of the approach presented here.