Previous studies of eyewitness testimony have found confidence and errors in testimony have an impact on credibility ratings and jurors' verdicts. The current study used the three factors of confidence, errors, and crime severity to determine their effects on witness credibility, verdicts, and guilt ratios. It was hypothesized that highly confident witnesses of a violent crime would be rated as more credible and result in more convictions compared to low confident witnesses. It was also expected that well calibrated witnesses of a nonviolent crime would results in higher credibility ratings and more guilty verdicts compared to poorly calibrated witnesses. Results showed significant differences between groups regarding credibility and guilt ratios. Witnesses of a violent crime were rated as more credible if they were highly confident, whereas witnesses of a nonviolent crime were rated as more credible if they did not make an error during testimony. Further, defendants on trial for a violent crime were rated as more guilty if witnesses were highly confident and did not make an error. The results do not support the effects of calibration, and instead offer support for the confidence-trumps hypothesis in violent crime situations. The current study also provides evidence that crime severity does influence jurors' ratings of credibility and guilt.