This study aimed to update normative data and establish cut-off scores for a fruit-based semantic verbal fluency (SVF) task among older Taiwanese adults as a method for detecting mild cognitive impairment (MCI). The task was chosen due to its familiarity and cultural neutrality for Mandarin-speaking populations.

SVF performance was evaluated in 245 healthy control participants and 360 individuals diagnosed with MCI. The influence of demographic variables was examined, and regression-based correction formulas were developed. Receiver operating characteristic (ROC) analyses determined optimal cut-off values according to established clinical classifications of MCI.

Age, education, and sex significantly influenced SVF performance. A demographically corrected 15th percentile threshold of 10 words was proposed for community screening. An optimal ROC-derived cut-off of 11.5 words yielded an AUC of .716 (95% CI: .68–.76), with sensitivity of 57.8% and specificity of 73.9%. SVF scores were significantly correlated with global cognition, memory, and processing speed.

The fruit-based SVF task is a quick, culturally relevant tool for detecting early cognitive impairment. Revised norms and cut-off scores can improve MCI identification in Mandarin-speaking seniors.