Research that undertakes longitudinal analysis of presidential discourse inevitably wrestles with a difficult issue: which presidential speeches should be analyzed? Surprisingly, there has been essentially no interrogation of this issue. Studies vary in the speeches they analyze: some focus on inaugural addresses, others on State of the Union addresses, yet others focus on some broader body of speeches, often with limited discussion of the criteria used for selection. The result is that when scholars set out to study broad trends in the rhetoric of America's most important political figure, they have little guidance in determining what speeches they should analyze. This article offers a detailed conception of major presidential addresses, argues that such addresses provide an ideal corpus of texts for longitudinal content analysis of modern presidential speeches, and briefly describes the contours of a data set consisting of 406 such speeches. This data set serves as a resource for scholars to rely upon when studying presidential discourse, one that might standardize future analyses so that more meaningful generalizations can be made and more precise replications can be undertaken.