A great deal of confusion has surrounded the basic definitions and concepts in the various versions of cross-impact analysis. The purpose of this article is to clarify the meaning of one of the fundamental concepts-conditional probability-as used in a cross-impact analysis. The authors begin by illustrating two versions of conditional probability, one based on correlation and one based on causation, and show that the latter is much better suited to the study of alternative futures. One of the main sources of past misunderstanding is the attempt to apply the correlative conditions of Bayes' theorem to a causative cross-impact analysis. They demonstrate that there is no inconsistency between Bayes' theorem and cross-impact analysis; the confusion results from the use of Bayes' theorem when the basic analysis involves causation.