Clinical-epidemiological research may orient us about the causes of disease, the relationships among them, and the relative magnitudes of their effects. The objective of this article is to link the notion of cause with the basic clinical-epidemiological parameters. There are different models explaining causality. All of them present the possible etiologic explanations for the diseases, taking into consideration the current knowledge at the time they have been posed. We start from a purely determinist conception, understanding causality as a constant connection between two factors x and y, unique, and perfectly predictable. Currently, this model is inadequate to be applied to many diseases. Many researchers have modified the determinist model to explain the multiple causality of disease, posing the existence of associations of causal factors, more than single factors, being these associations treated as sufficient cause (i.e. as a group of minimal conditions and events that inevitably produce the disease). That determinist concept of causality is supplemented with the probabilistic concept. The theory of probability is used in it, as well as the related statistical, methods, to empirically evaluate a possible association that is believed causal. As a consequence of the lack of certainty of the prediction at the individual level, the theoretical notion of cause is replaced by the empirical concept of risk factor, referring to a variable which is considered to be related to the probability that one individual develops the disease. Causal inference in epidemiology is the logic development of a theory, based on observations and arguments that attribute the presence (association) of a disease to one or more risk factors. We will follow the principles posed by B. Hill for the complex process called scientific generalization. To correctly perform this relationship between our ideas and are observations it is absolutely important to start from a correct election of the study design with which the research is undertaken.