The 59 college student subjects read court case summaries that included variable eyewitness evidence and constant alibi, circumstantial, and character evidence. In the first experiment, the frequency of guilty verdicts was significantly less when an eyewitness testified that the defendant was not the perpetrator, even when two other witnesses made positive identifications. In the second experiment, a low conviction rate was again associated with the presence of a nonidentifier, but only when the nonidentifier actually testified in court and stipulated that the defendant was 'not the man.' On the average, 70 percent of the jurors delivered guilty verdicts when both the victim and bystander rendered positive identifications; whereas, 12.5 percent delivered guilty verdicts when the bystander gave nonidentifying testimony. Guilty rates were unaffected by the identifying eyewitness' status and (in the first but not the second experiment) were higher when there were two identifying eyewitnesses. Tabular data and 31 references are provided. (Author abstract modified)