OBJECTIVES: This study aimed to evaluate the psychometric properties of four candidate cognition bolt-on items and their combinations to the EQ-5D-5L. METHODS: Four cognition items (concentration, memory, calculation, and learning) were developed as separate questionnaire items, and were administered with the standard EQ-5D-5L to 640 individuals in a general population survey in China. From the 4 items, 11 compound items were constructed, and the 'worse level counts' rule was used to calculate a compound item score. Psychometric performance of the cognition bolt-ons was assessed in terms of informativity, convergent validity, explanatory power, and discriminatory power. RESULTS: The tested four cognition bolt-on items improved the informativity, convergent validity, explanatory power, and discriminatory power of EQ-5D-5L, with calculation and learning yielding better psychometric performance. The compound bolt-on items that coverd a range of cognitive functions demonstrated superior psychometric performance compared to single-aspect bolt-on items, with those items covering calculation and learning resulting in better psychometric performance. CONCLUSION: This study confirmed the validity of the tested cognition bolt-ons in a general Chinese population. It supported the use of a compound bolt-on item covering a range of cognitive functions such as the ability to calculate and learn.