Wells and Lindsay (1983) proposed that juror belief of eyewitness testimony was influenced by various types of information. Four mock-jury experiments were conducted to test hypotheses derived from their metamemory analysis. In Experiment 1, 288 subjects read brief “transcripts” of a purse-snatching trial in which 0, 1, or 2 eyewitnesses testified for each of the prosecution and/or the defense. Consistent with the hypothesis that inter-subjective agreement (consensus among witnesses) was important, guilty verdicts were most likely in the presence of unopposed prosecution eyewitnesses and least likely in the presence of unopposed defense eyewitnesses. Experiment 2 employed 75 subjects viewing a videotaped trial simulation and replicated the findings from the first experiment as well as demonstrating that the nature of the defense witness' testimony (a no identification decision vs. alibi) was unimportant but the person providing an alibi was important (stranger vs. relative). In Experiment 3,60 subjects listened to an audiotaped trial procedure varying the internal consistency of the witness' testimony. Inconsistent testimony failed to reduce belief of the eyewitness as reflected in guilty votes. The fourth experiment exposed 60 subjects to audiotapes of a burglary trial varying lighting conditions and length of exposure of the criminal to the eyewitness. Neither variable significantly influenced belief of the eyewitness. Explanations for the failure to obtain significant effects in the latter two experiments and possible directions for further research are discussed.