Objective: To evaluate the false positive rate and false negative rate of the Chinese version of the 12-item General Health Questionnaire (GHQ-12) and the related factors in the epidemiological survey of mental ill-nesses in Zhejiang Province. Method: A total of 15000 subjects were randomly selected from the province-wide using multi-stage stratified cluster randomization. Analyses for this paper were made in the quality control sample,10% of the total 15000 subjects (1510 subjects) in which the Chinese version of Structured Clinical Interview for DSM-Ⅳ Axis Ⅰ Disorders (SCID) was used as a golden criterion, and a cutoff score of the GHQ-12 was set to ≥ 3 to define GHQ-12 cases. Results: Totally 1449 subjects (96.0%) completed both the GHQ-12 and the SCID. Adjusted for sampling effects, the false positive and negative rates of the GHQ-12 were respective 14.6% and7.8%. Adjusted for other considered correlates and sampling effects, the adjusted odds ratios (AORs) of GHQ-12 false positive with living in less economically developed urban areas (urban type Ⅱ), with self-reportedly poor physical health, with having ever sought help because of mental problems and with being currently married/co-habited were respective 2.23 (95% CI:1.24～4.01), 2.36(1.36～4.10), 1.53 (1.10～2.14) and 0.51 (0.30～0.86) while AORs of GHQ-12 false negative with being aged 35～49 year group and living in less econom-ically developed rural areas (rural type Ⅲ) were respective 2.59 (1.18～5.67) and 2.72 (1.21～6.14). Conclusion:Factors related to the GHQ-12 false positive and negative are different. The cutoff scores of the GHQ-12 should be used based on the characteristics of subjects during identifying or screening mental illnesses.