This paper asserts that reality testing is a complex ego activity which cannot be characterized globally as either intact or defective. In normals, neurotics, and "borderlines" it is actually a highly variable function. Some problems of nomenclature are addressed. Among many analysts there is an implicit tendency to concretize the means by which reality testing is performed. This may lead to certain conceptual problems and clinical inaccuracies. The relationship of reality testing to unconscious conflicts from all phases of development is emphasized and illustrated. Issues in the technical handling in analysis of manifestations of disturbed reality testing are discussed.