Since the inception of the modern atomic theory, chemists have used physical models to represent the structure of molecules. The goal of this paper is to bring molecular modelling into focus as a constitutive yet overlooked element of chemical practices. It begins with a short technical introduction to molecular models, and then moves into a participant-centred analysis of molecular modelling. Central points of this analysis include, first, a discussion of the dichotomy between graphical and material forms of representation, with suggestions about its consequences for a semiotically-centred view of scientific activity; and, second, a look at the problem of the interpretation of molecular models, as discussed in the chemical literature. The last section focuses on the design of modelling systems through two related historical case studies — namely, the production of two space-filling modelling kits developed in the United States between the late 1930s and the late 1960s.