Probabilistic scientific evidence is increasingly common in American courtrooms. Blood type, fiber, DNA, fingerprint, and hair sample evidence draw upon statistical data and probabilistic calculations. Although use of some of these forms of evidence remains controversial, in many contexts courts readily admit probabilities associated with that evidence. In criminal trials, however, courts are more reluctant to admit probabilistic calculations intended to identify the defendant as the wrongdoer. This study used 189 participants (83 men, 106 women) from the University of Minnesota to determine how mock jurors would manage probabilistic evidence. The seven independent variables were the strength of nonstatistical evidence, the quantification of nonstatistical evidence, the strength of statistical evidence, the combination of two pieces of statistical evidence, instruction in the use of Bayes' theorem, and presentation of fallacies (both prosecutor's and defense attorney's) concerning the use of statistical evidence. The subjects viewed one of 16 videotapes that presented a condensed mock trial. Subjects completed dependent measures after each of four witnesses and at the end of trial. The strength of both nonstatistical and probabilistic evidence affected verdicts; the other manipulations did not. Overall, subjects slightly underused the probabilistic evidence, as compared to their individualized Bayesian norms, and subjects did not succumb to fallacies; however, subjects greatly varied in overuse or underuse, even after Bayesian instruction. Future research should examine use of weak nonstatistical evidence and should test different probabilistic instructions. 5 tables, 4 figures, and 37 references