The application of inductive empirical variables in hegemonic alternatives of Epidemiology in order to treat socioeconomic variations in human groups has displayed certain limitations raising the need for new proposals. The use of social class was thus introduced as an analytical category operated according to place in the productive process for the subjects under study. This choice has already shown potentialities in several studies. This paper was intended to identify some limits and problems occurring when using the social class concept in epidemiological research. Besides the need for an adequate theoretical framework, construction of categories consistent with both their theories and the levels of abstraction employed, and the concern for more carefully analyzing the consumer and working modes for each social class (questions that have been discused elsewhere), this paper approaches several issues, as follows: the treatment of social classes as articulated clusters in the research process, the need to adjust the operational scheme to the particularities of each socioeconomic formation, the simplification resulting from the process of reducing the concept of social class to several linked variables, the need for a relatively large sample, the existence of debate over the definition of the position of class groups not participating in production, and subjects with two or more different positions in production.