In assessing the teaching practice of preservice teacher candidates, observers, particularly trained observers, will readily identify those who are exceptionally effective or ineffective. The anticipation is that university supervisors and mentor teachers will agree on who demonstrates extraordinary performance for a preservice candidate and who needs additional preparation before taking on solo classroom teaching responsibilities. The assumption is that candidates who exhibit outstanding skills in student teaching will excel on a teaching performance assessment and that those who fail the assessment will be those who struggle in student teaching. Identifying weak candidates is particularly critical to ensuring that beginning teachers do not earn licenses until they are competent and ready to teach full time. Both university supervisors' observations and teaching performance assessments aim to evaluate the competency of preservice teacher candidates, and both approaches prompt concerns among teacher educators about their use for licensing decisions. In an earlier study, the extent to which university supervisors' perspectives about candidates' performance corresponded with outcomes from a summative performance assessment (Sandholtz & Shea, 2012) was explored. The study specifically examined the relationship between supervisors' predictions and teacher candidates' performance on a summative assessment based on a capstone teaching event, part of the Performance Assessment for California Teachers (PACT). In contrast to expectations, researchers found that university supervisors' predictions of their candidates' performance did not closely match the PACT scores and that inaccurate predictions were split between over- and under predictions (for complete findings, see Sandholtz & Shea, 2012). The findings in that study, combined with suggestions from other researchers, prompted the authors to examine high and low performance through an in-depth follow-up analysis. In this follow-up study, the authors focus on four specific subsets of teacher candidates: not only the groups of high and low performers but also the groups of predicted-high and predicted-low performers, which were not examined in the earlier research. Analysis of the predicted-low performers (and within that group, the predicted-to-fail candidates) is important because that group includes candidates whom supervisors do not think are ready to be licensed yet pass the assessment. This follow-up study also expands the data sources and includes not only PACT score data but also information from student transcripts and student teaching. In addition, this study includes additional analyses that, for example, examine specific areas in the PACT to determine where differences occurred. The following questions are addressed: (1) Do academic background factors correspond with high or low performance on the PACT; (2) In what specific areas on the PACT do high- and low-performing candidates excel and fail; (3) To what extent do university supervisors accurately predict high and low performance on the PACT; and (4) To what extent do candidates whom university supervisors predict will fail the PACT end up passing? Data were drawn from candidates' records and included: (1) demographic and student teaching placement information; (2) student transcripts; (3) predicted scores for the PACT teaching event; and (4) actual scores on the PACT teaching event. The findings of this study highlight four issues related to the assessment of preservice teacher candidates. First, findings suggest that student teaching grades may not serve as discriminating forms of evaluation, even for candidates who perform particularly well or poorly on a teaching performance assessment. Second, the results of this study prompt questions about the connection between candidates' academic strengths and classroom teaching performance. Third, the lack of agreement in identifying exceptional candidates at both ends of the continuum warrants further investigation. Finally, the findings of the earlier study and this follow-up study raise questions about relying on a single measure to evaluate teacher candidates for licensing decisions.