Previous research suggests that appeals to anger and enthusiasm increase voter participation but decrease deliberation and openness to persuasion, while appeals to anxiety increase information-seeking and deliberation but not participation. Thus, campaigns have incentives to consider the emotional nature of their communications with voters and to differentiate distinct emotions that share valence categories. We examine the emotional content of emails sent by candidates for the US House to partisan supporters and test the effects of these appeals on voter behavior. Overall, emails contain more enthusiasm and anger relative to a set of political “letters to the editor” drawn from major US newspapers and, in some cases, relative to a set of randomly generated “pseudo-emails” based on word frequency in typical discourse. We do not, however, find any consistent differences in emotional language across candidate or campaign factors, such as incumbency and competitiveness. In a unique quasi-experimental study fielded during the 2018 election, using actual campaign emails, we also find little evidence that the emotional content of campaign communications influences citizens. This is true regardless of whether we measure emotional content using objective codings or subjective judgments by human coders. We discuss methodological difficulties encountered in our work and possible lessons for future research on this topic.