In this study, the authors evaluated the accuracy and precision of non-clinical testing for dental students' crown preparation proficiency on a manikin-mounted typodont as a predictor of performance on patients. Eighty-six seniors at Baylor College of Dentistry were evaluated by three calibrated senior faculty members from the general dentistry department. Four areas of interest were evaluated: occlusal reduction; axial reduction; margins, design, and finish line; and operative environment. Standardized grading scales were utilized to evaluate these preparation components. The kappa agreement statistic was used to assess agreement for the areas, which were then compiled to get an overall performance score. A Bland-Altman plot was generated and analyzed for agreement on overall performance between typodont and patient. Results of the analyses showed weak agreement between the two measurements for all four areas of interest with large variation in the differences. When the pairs of overall performance scores were compared, the Bland-Altman plot indicated that the typodont composite score was often larger than the patient composite score, especially for poorer performing students. The authors conclude that students' clinical performance on typodonts is a poor predictor of their clinical performance on patients since the students performed much worse on the clinical crown examination than on the typodont examination.