In this paper, I tried to evaluate the usefulness of intervening variable in the explanation or prediction of behavior. The usefulness of an intervening variable depends on the constancy of its relations to several independent variables and dependent variables for different conditions. From my point of view, the intervening variable must be a parameter between independent variables and dependent variables when mathematically formulated. In the Tolman-Hull method, a number of intervening variables are set up between stimulus variables (independent variables) and response variables (dependent variables); each is based on a single external or internal condition, and combined, they bring out the organism's behavior. In this method, an intervening variable x is finally determined by stimulus variables A as a function of A. Response variables B are deduced from x, and accordingly, they are also bound up with A. This relation is stated as a function of function as follows: B=f(x), x=f'(A). Concerning this relation, recent interpretation have pointed out that there is a distinction between hypothetical constructs and intervening variables. They presume that a hypothetical construct is a physiological model with physical properties of operation and an intervening variable an expediency in mathematical operation whose properties are neutral. They claim also that there is a discrepancy between abstruct levels and observable levels in these variables. According to the writer's proposition of intervening variables stated above, however, such a distinction or discrepancy does not appear. The relations of x to A and to B may be expressed in the following formulae: B=φf, A=ψ(x). They mean that dependent variables are not a function of function, and that an intervening variable is a parameter in its functional relations to independent and dependent variables. From this point of view, field theory, in which different events are put on a same level, is valuable to psychological studies of behavior.