Background: Problem-based learning (PBL) allows faculty to observe students interact and solve problems. Thus, it represents a prime opportunity to provide authentic feedback on learners' knowledge, skills, and attitudes. However, we are concerned that PBL faculty do not accurately convey feedback to students. Methods: To assess the difference between formal evaluations and candid assessments of student performance, we conducted a study of 178 preclinical medical students at Kuwait University. We quantitatively compared PBL evaluations of students with candid assessments of students' competence as obtained from structured interviews with 19 PBL facilitators. We also compared facilitators' comments on the module evaluations with candid comments solicited during the interviews. Results: We did not find a strong quantitative or qualitative correlation between faculty feedback and their candid impressions of student performance. Thematic analysis of the comments disclosed multiple factors that influenced the accuracy and specificity of faculty feedback. Conclusions: Systematic discrepancies between feedback given to students and actual assessments of their performance can result in false reassurance of competence, which undermines our curricular efforts and prevents the trainee from achieving his or her full potential.