Although normative political theorists have argued that citizens talking with other citizens about public affairs is essential in a democracy, empirically oriented political scientists have tended to ignore political discussions. This paper draws on National Election Studies and General Social Surveys to plumb the extent and breadth of political conversations in the U.S. We also explore who talks about public affairs, with whom they speak, and why some people avoid discussing politics. A logit analysis of pooled National Election Studies from 1984 to 1992 shows that some factors typically thought to affect taking part in politics are not significant predictors of the probability that people will engage in political discussions, but other variables not normally included in participation models do affect talking about politics.