Explicates and evaluates 5 distinct versions of the 2-factor theory of job satisfaction. It is concluded that (a) 2 of these versions are invalid, as they are not supported by any empirical studies; (b) another version is invalid, as its alleged empirical support merely reflects E coding biases; and (c) the validities of the remaining 2 versions are, at present, indeterminate, as they have not been adequately tested in studies where defensive biases inherent in certain self-report measures are eliminated. These conclusions are discussed in terms of the principle of multiple operationalism, and a method of testing the 2 indeterminate versions is suggested.