The explicit solution of the problem of maximization of information divergence from the family of multinomial distributions is presented.