This paper presents a discussion of the use of educational tests in guidance services as seen in the light of modern developments in statistical theory and computer technology, and of the increasing demands for such services. A focus and vocabulary for this discussion is found in Turnbull's recent article on "Relevance in Testing." Following an introductory discussion of the need for guidance services some very recent work in Bayesian inference is reviewed and the implications of this work for educational research methodology are noted. Special attention is given to the Lindley equations which provide solutions for a number of problems in the comparative prediction of academic achievement. The suggestion here is that in a changing educational environment the Bayesian methodology can provide an increase in the effectiveness and applicability of such programs as Horst's monumental Washington Pre-College Testing Program. Comparative prediction is seen as an idea whose time has come. (Author)