ABSTRACTThis paper re‐examines Duverger's hypothesis about the ‘multiplicative tendency’ of the electoral system of proportional representation. The data cover six countries over time, and the change of the electoral system to PR is conceived as an ‘intervention’ whose impact on the party system is assessed by time‐series techniques following the logic of ‘interrupted time‐series quasi‐experiments’. In two countries the results are altogether negative, in one positive. In the three other cases, PR does not increase the degree to which the party system is fractionalized, but increases the number of parties, essentially by adding small ones. The support for the hypothesis is at most partial and, in addition, it is argued that the observed effect is systemically unimportant.