Abstract Development of specific constitutive equations (both mechanical and thermodynamic) has been carried out along three different of approach. The oldest one dates back to the second half of the eighteenth century. It is basically a trial-and-error procedure, and it has produced most of the currently available constitutive equations. Enormously successful as this approach has been in its first steps, it appears to become hopelessly unsuccessful with increasing complexity of rheological behavior. The second procedure, which dates back to the early 1940's, has had considerable impetus in the last ten years. It is grounded in molecular modeling; it has definite advantages over the first procedure; it has also been remarkably successful. However, this procedure may also be running close to its limit of usefulness. A third procedure can be identified in a few recently published works. In this paper, after a critical review of the first two procedures, the potentialities of the third procedure are evaluated.