The present study examined the influence that a juror's first impressions of an expert witness might have on two outcomes: judgments of the witness' credibility, and verdict decisions in a criminal case involving a Not Guilty by Insanity (NGRI) defense. This was the first study to use "thin slice" methodology to manipulate time exposed to expert testimony and assess reliability of witness credibility ratings over time. This study also examined the degree to which these impressions influence the relationship between juror opinions and jury decision-making. A 2 (non-deliberating vs. deliberating jury) X 3 (observing 30 seconds, 5 minutes, or 10 minutes of expert witness testimony) between subjects design was implemented. Participants (N = 188, 30 mock juries) viewed a videotaped presentation of testimony from an actor portraying a forensic mental health professional called on by the defense. Mock juror characteristics, responses to a thought listing measure, and transcriptions from the videotaped jury deliberations were coded for exploratory analysis. Primary results, obtained via Hierarchical Linear Mixed Modeling to account for the random effect of group, were supported by jury-level analysis. Despite support for the accuracy of "thin slice" judgments in the literature, results found that jurors in the 30 second condition judged the expert as significantly less credible in this study. Results did not support the anticipated leniency shift in juries post-deliberation, and instead, yielded a significant two-way interaction on verdict for the 30 second group, such that non-deliberating jurors were more lenient than deliberating jurors. Implications for understanding how impressions of expert witness testimony translate from the juror to the deliberation room are discussed, with particular attention to cases with an increased likelihood of bias against the NGRI defense.