With the corpus organized in the way described in the first part of this article (JIPA 4.2, December 1974) it was possible to move on to the main objective, namely the extraction of a system of intonation from the corpus. In the present work it was a fundamental requirement of the eventual system of intonation that it should account for all the intonation patterns observable in the corpus. It is suggested that this system provides a very adequate means of describing the intonation of modern colloquial Russian in the broader context, although it is not, of course, possible within the limits of this article to test that proposition. If the system is to be descriptive of the intonation of Russian in the widest sense it seems essential that as few assumptions as possible should be made about what is and what is not functional in an intonation system. The problems involved in setting up and testing hypotheses about the function of intonation are considerable, and the only assumption made in this work is that perceivably different intonation patterns could have different functions in the conversational situation—the definition of those functions is not a necessary part of the task here. Given this approach it follows that an intonation system is not regarded as comparable to a phonemic system, in which a relatively small set of functional units will be realized in a very large number of different but not necessarily contrastive ways. Rather it is viewed here as a very much larger set, graded in terms of probability of occurrence.