We show that, using a Support Vector Machine classifier, it is possible to determine with a 75% success rate who dominated a particular meeting on the basis of a few basic features. We discuss the corpus we have used, the way we had people judge dominance and the features that were used.