At the present stage of the so called structuralist approach to the philosophy of science, there seems to be some difficulties which are purely logical. Since this approach has an increasing influence on philosophers and it seems to be potentially very fertile, it may be important to try to remove them. I have already criticized some of them in [7], but here I will make some positive suggestions which are more detailed than in that paper. By the structuralist (or set-theoretical) philosophy of science I mean, ac? cording to the current terminology, the approach whose basic principles can be found in Sneed [10], Stegm?ller [12], etc. It will be assumed here that the reader is familiar with it. These logical difficulties can be divided into two general groups. First, the structuralists claim that they use intuitive (naive) set theory as the framework in characterizing the logical structure and dynamics of empirical theories. However, intuitive set theory has to be used cautiously, as we have been taught by the set-theorists of this century. Thus the following problems can? not be ignored in the foundational studies of set theory: What kinds of entities can be intuitively regarded as sets? What set-theoretical operations can be consistently applied to different entities occurring in set theory? Now it seems that the structuralists have ignored (deliberately?) the advice of the set-theorists in these problems, with the result that their system is inconsis? tent. The basic set-theoretical trouble in the structuralist approach derives, roughly, from the fact that it attempts to replace sentences by collections of models without noticing that it can mean a change of entities which are sets for entities which are not. This does not mean of course that the structuralist system would not be useful. But it could be better to use intuitive set theory so that it would not lead into perspicuous contradictions and would better correspond to our more or less trained intuitions. Therefore, I will make some suggestions that will maintain, as I hope, the main lines of the structuralist approach. The second difficulty is related to model theory. The structuralists operate with concepts whose analogs can be found in general model theory. However,