The least square error (LSE) method is frequently used in determination of Morison’s coefficients from the analysis of measured wave forces on circular cylinders. It will be proved in this paper that Morison’s coefficients determined from the LSE method always lead to predicted forces which have smaller variances (2nd-order statistical moments) than those of the measured forces. In other words, it is proved that predicted forces from this method are of lower intensity compared to the measured forces, and hence can be unconservative. A new method, termed the maximum correlation (MC) method, has been introduced. This method has all the advantages of the LSE method, but leads to predicted forces whose variances are equal to those of the measured forces.