A total of 120 male and 168 female students at a junior college were paid $5 each to take part in the 2-hour study. These mock jurors watched video tapes of a realistic simulated trial. They rendered verdicts individually before deliberation and as groups after deliberation in six-member groups. The simulated trial was a felony-murder case in which the defendant was alleged to have robbed a liquor store and murdered the female proprietor. The trial contained inadmissible evidence supporting conviction, inadmissible evidence supporting acquittal, or no inadmissible evidence. The judge's instructions emphasized the importance either of protecting the defendant's due process rights or of reaching an accurate verdict. Results showed that the variation in judges' instructions had no effect. However, jurors who received inadmissible evidence supporting acquittal were less likely to convict than were those who received either proconviction inadmissible evidence or no inadmissible evidence. The conviction rates in the latter two conditions did not differ significantly. Paradoxically, jurors who received inadmissible evidence supporting conviction thought that their verdicts had been influenced by it, whereas those who received inadmissible evidence supporting acquittal thought that their verdicts had not been influenced by it. The presence of inadmissible evidence did not affect jurors' evaluations of the strength of admissible evidence. The reason for asymmetry regarding effects of evidence supporting conviction and evidence supporting acquittal was not known. This gap in understanding and the study's reliance on mock jurors point to the need for caution in generalizing the study's findings. Tables, reference notes, and 20 references are included.