The process of model building is a basic activity for any scientific discipline, and sociology is no exception to this rule. The sociological literature devoted to clarify the epistemological and methodological assumptions behind each modelling technique is rather scarce, however. This article intends to provide some ideas on this topic. In order to do so, it tackles onto the scientific status of modelling techniques in sociology, especially those expressed in artificial or formal languages. In addition, it stresses the importance of these techniques in building or validating causal explanations, and of the role played by social theory in the specification of such models. The three most common and widely used modelling techniques in sociological research are considered, namely, statistical modelling, mathematical modelling and social simulation. The article emphasises the theoretical underpinnings of these modelling techniques, the kind of explanation we can draw from them, and the main drawbacks, both practical and substantive, they present for social research.