AbstractIn the present paper the reliability and predictive capacity of multiple‐choice, essay, and viva voce (oral) examinations in anatomy are compared. In an attempt to improve the reliability of the viva, a structured rating system was designed. In the latter system candidates were rated separately for achievement under the headings of Factual Recall, Analysis and Interpretation of Data, Problem Solving, and Verbal Presentation. The predictive capacity of examinations in anatomy was estimated by calculating the correlation between scores in anatomy and the score obtained by the same students in final‐year medicine. The reliability coefficients of the examinations were estimated using the split‐halves technique, together with Spearman‐Brown formula in the case of the viva and the essay exam, and using the Kuder‐Richardson formula 20 for the multiple‐choice tests. The correlations between scores in anatomy examinations and final‐year medicine examinations show that scores from essay and viva voce examinations are better predictors of success later in the medical course than are scores from multiple‐choice examinations. The reliability coefficient of viva voce ( < 0.4) examinations is clearly less than that of either essay examinations (around 0.65) or multiple‐choice tests (usually > 0.8). However, the use of the structured rating procedure elevated the reliability coefficient of the viva voce examination from between 0.3 and 0.4 in the case of the unstructured examination to between 0.7 and 0.8 in the case of the structured examination.