This study examines a debate among experts sponsored by Cornell University in 2014 on whether or not to allow hydrofracking in New York State. The focus is on the question-answer portion of the debate to see how risk is discursively constructed from experts' claims and rejoinders as well as audience participation. The granular methodology of discursive analysis is used to examine how risk gets talked into being and amplified or mitigated through interaction in the question-answer portion of the debate. Risk gets constructed through participants' practices of metadiscourse—how they formulate what has been said, report the speech or actions of others, or repeat certain locutions into lists for rhetorical effect. These metadiscourse practices provide a resource for the debater to critically characterize other's words or deeds prior to presenting their preferred position.