ABSTRACTThe application of a model of Communist voting to two rather different geographical and historical contexts yielded varying empirical results. The model, composed of traditional socioeconomic explanatory variables as well as organizational and historical variables, proved fruitful in the Norwegian case, much less so in the Catalan case. This difference in explanatory power may be understood in terms of varying levels of modernization. Models of electoral behaviour should be more sensitive to developments of this kind. Higher flexibility in the use of models is thus desirable. What is needed is a battery of models of electoral behaviour, and a more conscious attitude on the part of electoral sociologists to the implications of time and space for the choice of theoretical approach.