This review of the relevant literature in the area of numerical facility attempts to clarify the construct of numerical facility and provide guidance for items tapping this ability. The review is presented in five parts. The first section introduces two approaches that can be used to investigate numerical facility, including factor analysis. Studies that have attempted to isolate a numerical factor are presented in the second section. The third section deals with mathematical ability, a construct typically viewed as related to numerical facility. Several studies relating numerical facility and "automatic processes" are presented in the fourth section. The fifth section supports Thurstone's (1938) conclusion that the ability to manipulate numbers is a unique ability and that there is some evidence that this is genetic in nature. Although most of the studies measuring this ability used problems involving simple arithmetic, this may not be the best way to measure numerical facility. It is suggested that tasks involving the manipulation of symbols other than numerals might be preferred. A recommendation is made that the Johnson O'Connor Research Foundation use three tests to research this construct: (1) simple arithmetic; (2) simple arithmetic involving some arithmetic reasoning; and (3) nonnumerical manipulation. (KSA)