While there is extensive research on the language of twitter, our knowledge of the pragmatics of particular twitter genres (and sub-genres) is still piecemeal. At the same time, in the past decades, political discourse analysis has widened our understanding of how language can be used instrumentally to alter or manipulate public interaction, meanings and opinions. However, it has seldom examined the evaluative load of political communication in much detail. To this end, the paper, on the one hand, serves to illuminate the pragmatics of political tweets as a twitter genre. On the other hand, the study brings to the fore the strategic use of negative evaluations in political online campaigning and discusses its potential (and actual) socio-political ramifications. The quantitative and qualitative analysis of negative evaluations largely draws on Martin and White’s Appraisal framework (2005) and is based on a compatible study by Cabrejas-Peñuelas and Díez-Prados (2014). The paper tracks down, classifies and categorizes the negative evaluations of a subset of twitter posts by Donald Trump and Hillary Clinton in a self-compiled corpus of 1965 tweets, with a view to evaluation types, their relative frequencies and dispersion across the corpus, as well as objects and targets of evaluation. The quantitative analysis is then completed by a qualitative examination of the objects and targets of evaluation in both twitter profiles as well as a closer look at the recurrent language used to evaluate the political “other”. The results show that Trump makes more flexible (and strategic) use of negative evaluations (both in terms of types, frequency and distribution), while Clinton’s negative evaluations are less frequent, less diverse and, thus possibly, less convincing. Keywords: evaluation, corpus pragmatics, Appraisal Theory, political discourse analysis, Twitter