Telepresence means business people can make deals in other countries, doctors can give remote medical advice, and soldiers can rescue someone from thousands of miles away. When interaction is mediated, people are removed from and lack context about the person they are making decisions about. In this paper, we explore the impact of technological mediation on risk and dehumanization in decision-making. We conducted a laboratory experiment involving medical treatment decisions. The results suggest that technological mediation influences decision making, but its influence depends on an individual's self-construal: participants who saw themselves as defined through their relationships (interdependent self-construal) recommended riskier and more painful treatments in video conferencing than when face-to-face. We discuss implications of our results for theory and future research.