Previous research has not supported the validity of Olson's “curvilinear hypothesis.” Convergent validity studies of the FACES I and II have indicated that the “cohesion” and “adaptability” dimensions have a linear relation with healthy family functioning. It is not clear, however, whether these findings are due to the measurement of the FACES' constructs or to the curvilinear theory. This study tested the “curvilinear hypothesis” in terms of these theoretical and measurement issues. The results indicate that FACES II and III do not adequately operation-alize the “curvilinear hyothesis.” However, an item consistent with the “curvilinear hypothesis” utilizing a bipolar item response format did adequately opera-tionalize the theory. The use of a bipolar item response format is suggested in the measurement of “cohesion” and “adaptability.”