We show that overconfident individuals are likely to be arrested for public intoxication by using arrest records from a university town police log. This relationship is robust to various control variables such as risk aversion, time discounting, present bias, self-control, selfishness, loss aversion, and socializing with peers arrested for public intoxication. However, this relationship is no longer significant using only self-reported arrest data. We hypothesize that overconfident individuals are likely to underreport their arrests. This result has important implications for the use of self-reported data on public intoxication arrests rather than actual arrest records.