Deterrence theorists and researchers have argued that the critical dimension of sanction certainty is its level-increasing the certainty of punishment from a lower to a higher level will inhibit criminal conduct. However, the true certainty of punishment is rarely known with much precision. Both Sherman (1990) and Nagin (1998) have suggested that ambiguity about the level of punishment certainty is itself consequential in the decision to commit or refrain from crime. Here, we investigate this proposition. We find some evidence that individuals are "ambiguity averse" for decisions involving losses such as criminal punishments. This finding means that a more ambiguous perceived certainty of punishment is a greater deterrent of some crimes than a nominally equivalent but less ambiguous one. However, this effect depends on how large an individual's risk certainty perception is initially. That is, we find evidence for "boundary effects" (Casey and Scholz, 1991a, 1991b) in which this effect holds for lower probabilities but reverses for higher ones. For higher detection probabilities, individuals become "ambiguity seeking" such that a less ambiguous detection probability has more deterrent value than a nominally equivalent but more ambiguous detection probability. Results are presented from two distinct, but complementary, analysis samples and empirical approaches. These samples include a survey to college students with several hypothetical choice problems and data from the Pathways to Desistance study, a longitudinal investigation of serious adolescent offenders transitioning from adolescence to young adulthood.