Objective(s) To determine the normative cut-off value for the free androgen index (FAI) among reproductive-age women in China. Design Cross-sectional, population-based study. Setting A multistage systematic cluster random sampling among 16 communities from two rural and two city regions. Patient(s) A total of 1526 women were randomly selected from a representative sample of 3009 women aged 20-45years from the general population of southern China. Among these women, 444 subjects without known factors affecting androgen levels were selected as the reference group. Intervention(s) Subjects underwent physical examination, blood tests of hormone and metabolic profiles, and ultrasound evaluations. Measurement of the main outcomes The FAI was calculated according to the following equation: FAI=100 x total testosterone (nmol/l)/sex hormone-binding globulin (nmol/l). The normative cut-off value of the FAI was computed based on the 5th95th percentiles and K-means cluster analysis (K=2). Results A reference range of 0.76.4 of the FAI was indicated by the 5th and 95th percentiles. K-means cluster analysis identified 6.1 as the cut-off value for the FAI in the total population, which was close to the upper limit determined by the 95th percentile. Hyperandrogenic women present with elevated risk factors for metabolic disorders. The mean FAI values decreased with age among the total population, as evaluated by either K-means clustering or a percentile analysis. The FAI value was negatively correlated with age (r=-0.185, P<0.0001) and positively correlated with body mass index after adjustment for age (r=0.250, P<0.0001). Conclusion A reference range of 0.76.4 FAI was indicated by the 5th and 95th percentiles among the reference group, which excluded the subjects with factors known to affect androgen levels; 6.1 was identified by the K-means cluster analysis as the cut-off value of the FAI for the general population. Hyperandrogenic women present with higher risks for metabolic disorders.