The method of mathematical model self-organization is applied to handle observations of the hydrological, chemical and biological parameters of the Sea of Azov, conducted from 1926 through to 1985, in order to identify model equations of the latter's ecosystem. A set of 43 evolutionary equations have been employed to generate a picture of the ecosystem parameters' continuous variability and to identify relationships between them. Model calculations of the total catch of freshwater fishes, primary productivity, zoobenthos, phytoplankton, and other characteristics have demonstrated their being in good agreement with the observations conducted from 1950 to 1985. A prediction of the ecosystem parameters' evolution for the period from 1985 to 1995 has been made.