The quality of surface water plays a vital role in determining the sustainability of the ecological environment, the health of the population, and the socio-economic development of entire countries. Unfortunately, the rapid growth of the world's population together with the current climate change mainly deteriorates the state of surface water bodies. Thus, the use of effective methodologies capable of quickly and easily obtaining reliable information about the quality of surface water becomes fundamental for the effective use of water resources and implementation of mitigation measures and actions. Water pollution indices are one of the most widely used methods for providing a clear and complete picture of the state of river pollution, for the needs of rational water use and sustainable management of water resources. The selection of parameters is one of the most important and difficult stages, and the available statistical methods do not demonstrate great objectivity and accuracy in determining the real state of water quality. a new approach, based on the theory of entropy and known as the principle of maximum informativeness with minimum redundancy of information (MIMH), is proposed for determining the optimal subset of parameters describing the change in the quality level of a water body in space and time and, thus, determining the sources of pollution. The algorithm for the MIMN principle was implemented and applied to three rivers: the Southern Bug, the Dniester, and the Siverskyi Donets.