Waves statistics are important for the evaluation of a service performance of a ship. In general, waves statistics are given as a scattering table of significant wave height and mean wave period based on observation results. To make it easy to treat the table, researchers have proposed analytical distribution of wave statistics. They have been often applied by lognormal distribution and Weibull distribution. However, theoretical grounds of lognormal distribution and Weibull distribution are not so clear. Authors propose new stochastic models of wave statistics based on the regression to equilibrium condition. The models are expressed by stochastic differential equation with Gaussian process, and the distribution of wave statistics are obtained theoretically. The validity of the distribution is evaluated by the observation results of significant wave height and mean wave period. In addition, the time series simulation of significant wave height and mean wave period is shown.