This paper is the continuation of a preceding one which appeared in this journal, and presented an approach to self-organizing systems based upon information theory. To begin with, some illustrating examples try to show why one can think that the concept of information should play a crucial role in a theory of general systems. Then the concept of ‘ structural entropy ’, and the ‘ evolution principle ’ that we have previously introduced, are improved here in a definitive form. The concept of ‘ information potential ’ is defined and we show that the capacity of an information source is given by the logarithm of its inverse entropy rather than by its negative entropy. Composition laws for the structural entropy are stated and it is shown that these concepts provide an approach to information balance. Some simple examples are outlined to connect this approach with well known existing results.