In this paper, a new theoretical method to estimate the ship speed in actual seas using probabilistic method is proposed. The speed loss is classified into the following five factors: (1)aging effect, (2)fouling effect, (3)wave effect, (4)wind effect and (5)ocean current effect. These effects are described with probability density functions. As all factors are independent of each other, the convolution integral is available for obtaining the joint probability density function of them. A container liner is selected for the calculation and her service route is Pacific South-West Coast route (PSW). A long term prediction for 10 years is conducted using wind and wave statistics of the route.