Past confusion over the concept of projection, especially with regard to the related research evidence, has stemmed from the fact that numerous types of projections have been proposed which were not clearly differentiated from each other. In the present paper it is pointed out that theories of projection differ with regard to 2 major points: (a) whether the individual projects his own trait or a different (complementary) trait, and (b) whether the individual is aware or unaware of possessing the trait which results in the projection. The 4 possible combinations resulting from these theoretical differences are plotted in a 2 X 2 table, which provides a system in which the various types of projection and related research can be classified and compared. A critical review of the research on each type reveals that there is strong evidence for the projection of S's own trait or the complement of this trait if S is aware that he possesses the trait. There is no evidence for any type of projection resulting from a trait which S is not aware that he possesses. This finding calls into question the existence and interpretive use of "classical" (or similarity) projection, the type of response referred to most frequently by the term "projection.