This article examines data drawn from a 2001 Ontario (Canada) provincial inquiry into the deaths of seven people as a result of water contamination in a small Ontario town. The examination focuses on question-answer sequences in which the premier of Ontario, Michael Harris, attempted to resist lawyers' attempts to control and restrict his responses. In particular, on the basis of the data it is argued that the power of cross-examining lawyers does not reside solely in their ability to ask controlling and restrictive questions of witnesses, but rather is crucially dependent on their ability to compel witnesses to produce straightforward, or “type-conforming,” answers to these controlling and restrictive questions. The witness whose testimony is analyzed was not compelled to produce answers that logically conformed to the form of the lawyers' questions (i.e., “yes” or “no”) and, as a result, often usurped control over the topical agenda of the proceedings. In this sense, the present work builds on Eades's conclusion that “we cannot rely on question form to discover how witnesses are controlled.”A previous version of this article was presented at Sociolinguistics Symposium 15, Newcastle, U.K., in April 2004. We thank audience members at that conference and two reviewers for Language in Society for comments on previous versions. The research on which this article is based was funded in part by a SSHRCC Regular Research Grant (#410-2000-1330) to the first author.