Purpose: The didactic lecture remains one of the most popular teaching formats in medical education; yet, factors that most influence lecturing success in radiology education are unknown. The purpose of this study is to identify patterns of narrative student feedback that are associated with relatively higher and lower evaluation scores.Methods: All student evaluations from our core radiology elective during 1 year were compiled. All evaluation comments were tagged, to identify discrete descriptive concepts. Correlation coefficients were calculated, for each tag with mean evaluation scores. Tags that were the most strongly associated with the highest-versus lowest-rated (> or < 1 SD) lectures were identified.Results: A total of 3,262 comments, on 273 lectures, rated by 77 senior medical students, were analyzed. The mean lecture score was 8.96 AE 0.62. Three tags were significantly positively correlated with lecture score: "interactive"; "fun/engaging"; and "practical/ important content" (r ¼ 0.39, r ¼ 0.34, and r ¼ 0.32, respectively; all P < .001). More tags (n ¼ 12) were significantly negatively correlated with score; the three tags with the strongest such correlation were: "not interactive"; "poorly structured or unevenly paced"; and "content too detailed or abundant" (r ¼ e0.44, r ¼ e0.39, and r ¼ e0.36, respectively; all P < .001). Analysis of only the highest-and lowest-rated lectures yielded similar results.Conclusions: Several factors were identified that were strongly associated with lecture score. Among the actionable characteristics, interactive lectures with appropriately targeted content (ie, practical/useful) were the most highly rated.