Two studies test whether people believe in optimal deterrence. The first provides people with personal injury cases that are identical except for variations in the probability of detection and explores whether lower probability cases produce higher punitive damage awards and whether higher probability cases produce lower awards. No such effect is observed. The second asks people whether they agree or disagree with administrative and judicial policies that increase penalties when the probability of detection is low and decrease penalties when the probability of detection is high. Substantial majorities reject these administrative and judicial policies. Policy implications for the role of the jury in achieving deterrence are explored. Copyright 2000 by the University of Chicago.