The direct parameter estimation algorithm is applied to the simple equation of rolling motions. The algorithm is composed of the IIR (Infinite Impulse Response) digital filters and RLS (Recursive Least Squares) algorithm. The IIR digital filters are introduced to avoid taking direct derivatives and to reduce the effects of measurement noise. Furthermore, the digital filters are free from the initial condition problem and the accumulation problem of integration error. In order to discuss the quantitative accuracy, the authors carried out model experiments. At the first, the present method is applied to the time history data of free rolling motion tests in the still water. Next, the method is applied to the random rolling motion data in irregular waves. In both cases, there is satisfactory agreement of the dynamic rolling motion parameters identified by the present method and the previously measured parameters. This concludes that the present method is a powerful tool for on-line identification of dynamic rolling motion parameters. Finally, in order to verify the practical usefulness, we also carried out on board tests. The results of these tests are shown and problems encountered on applying are reported.