The problem of violence within America's prison system is not only misunderstood in key respects by the penological science that seeks to know it; it is in significant respects an effect of penological science and the practices it generates in the course of having aligned itself with legal discourses of the political state. Our thesis, informed by concepts that are central to Michel Foucault's antiscience project, is that violence within America's prison system is best understood as an oppositional response to a penological science, which, organized as a set of discursive practices and deployed in an alliance with legal discourses of the political state, succeeds in knowing inmates only in restricted and ultimately self-defeating ways. Our thesis is neutral as to the truth or falsity of the specific knowledge claims of penology. Rather our focus is on the ways in which retrictedness and self-defeat are inherent in the very doing of penology and how, further, the purported successes of penology's discursively grounded knowledge claims, as well as the practices they engender, contribute to the (violent) undoing of those very same claims. W e offer illustrative empirical support for our thesis by showing some discursive practices of penological science, how they are deployed specifically in the course of inscribing inmates as scientifically knowable objects, and how such tactics incite inmates to develop new identity resources, steeped in violence, which effectively subvert penology's claims to knowledge.