We report results from economic experiments of decisions that are best described as petty larceny, with high school and college students who can anonymously steal real money from each other. Our design allows exogenous variation in the rewards of crime, and the penalty and probability of detection. We find that the probability of stealing is increasing in the amount of money that can be stolen, and that it is decreasing in the probability of getting caught and in the penalty for getting caught. Furthermore, the impact of the certainty of getting caught is larger when the penalty is bigger, and the impact of the penalty is bigger when the probability of getting caught is larger.