When students encounter a set of concepts (or terms or principles) that are similar in some way, they often confuse one with another. For instance, they might mistake one word for another word with a similar spelling (e.g., allusion instead of illusion) or choose the wrong strategy for a mathematics problem because it resembles a different kind of problem. By one proposition explored in this review, these kinds of errors occur more frequently when all exposures to one of the concepts are grouped together. For instance, in most middle school science texts, the questions in each assignment are devoted to the same concept, and this "blocking" of exposures ensures that students need not learn to distinguish between two similar concepts. In an alternative approach described in this review, exposures to each concept are "interleaved" with exposures to other concepts, so that a question on one concept is followed by a question on a different concept. In a number of experiments that have compared interleaving and blocking, interleaving produced better scores on final tests of learning. The evidence is limited, though, and ecologically-valid studies are needed. Still, a prudent reading of the data suggests that at least a portion of the exposures should be interleaved. (Contains 4 figures.)