Episodic memory starts to decline very early in the development of Alzheimer’s disease (AD). Subtle impairments in memory binding may be detected in mild cognitive impairment (MCI). This study aims to examine the psychometric properties of the Chinese version of the memory binding test (MBT).

One hundred and sixty-four subjects (26 individuals with AD, 67 individuals with amnestic MCI (aMCI), 30 individuals with subjective cognitive impairment (SCI), and 41 cognitively normal elderly individuals (NC)) participated in the study. Twenty-two subjects repeated the assessment of the MBT within 6 weeks (± 2 weeks). Pearson correlation was used to calculate the convergent validity. The test––retest reliability was determined by the calculation of the intraclass correlation coefficient (ICC). Discriminative validity was calculated to evaluate the receiver–operating characteristic curves. The optimal index was chosen by comparing the area under the curve for specificity and sensitivity ≥ 0.80. The optimal cutoff score of the index was chosen to maximize the sum of sensitivity and specificity.

The absolute value of the convergent validity of the direct indexes of MBT ranged from 0.443 to 0.684. The ICC for each of direct indexes was 0.887–0.958. Total delayed paired recall (TDPR) was the optimal index for discriminating aMCI from NC. The cutoff score for TDPR was ≤25 to distinguish aMCI from NC (sensitivity = 0.896, specificity = 0.707).

The Chinese version of MBT is a valid and reliable instrument to detect MCI.