The two most common types of electoral systems (ES) used in electing national legislatures are proportional representation and plurality voting. When they are evaluated, most often the arguments come from social choice theory and political sciences. The former overall uses an axiomatic approach including a list of mathematical criteria a system should fulfill. The latter predominantly focuses on the trade-off between proportionality of apportionment and governability. However, there is no consensus on the best ES, nor on the set of indexes and measures that would be the most important in such assessment. Moreover, the ongoing debate about the fairness of national elections neglects the study of their vulnerabilities. Here we address this research gap with a framework that can measure electoral systems' vulnerability to different means of influence. Using in silico analysis we show that plurality voting systems are less stable than proportional representation. They are also more susceptible to political agitators and media propaganda. A review of real-world ES reveals possible improvements in their design leading to lower susceptibility. Additionally, our simulation framework allows computation of popular indexes, as the Gallagher index or the effective number of parties, in different scenarios. Our work provides a new tool for dealing with modern threats to democracy that could destabilize voting processes. Furthermore, our results add an important argument in a long-standing discussion on evaluation of ES.