Citizens express attitudes toward politicians and policies as special kinds of judgments colored by emotions and we call them sentiments. In this way, they participate in the process of creating public opinion, which can be manifested in different ways. Nowadays, it could be expressed through citizen comments on Websites, such as news portals, blogs and forums. Sentiment analysis of comments can give us an insight into the fundamental beliefs of citizens and their political profi les, which is very important for politicians when they want to explore public opinion, especially at the time of elections. This method is based on the extraction of sentiments expressed by words and terms in citizen comments on the Web. This paper presents the results of automated or manual sentiment analysis of citizens’ comments about politics on news websites in different countries. It shows that citizens generally have a dominantly negative public opinion on politicians and politics and that different news websites gather specifi c political profi les of citizens. The comparison between the characteristics of public opinion and election results suggests that the results of sentiment analysis can serve as predictors of election success.