Forced-choice (FC) measures are becoming increasingly common in the assessment of personality for high-stakes testing purposes in both educational and organizational settings. Despite this, there has been relatively little research into the reliability of scores obtained from these measures, particularly when administered as a computerized adaptive test (CAT). This study examined the test–retest reliability of an FC personality CAT, comparing its reliability to the reliability of personality measures using a Likert-type rating scale. Using a relatively large sample (N = 743), participants completed multiple personality measures across two time points. The test–retest reliability estimates for the personality dimensions had a mean of .63 with a mean reliability of .73 when formed into Big Five personality trait composites. The FC personality reliabilities were lower than those of the Likert-type scales, though the findings are within the range of those found in meta-analytic studies.