Purpose: Despite the wide usage of World Health Organization Disability Assessment Schedule II (WHODAS 2.0) in psychiatry research and clinical practice, there was limited knowledge on its proxy reliability among people with mental disorders. This paper aimed to compare the 12-item WHODAS 2.0 responses of adult patients with mental disorders to their family caregivers. Methods: In this study, 205 pairs of patients with mental disorders and primary family caregivers were consecutively recruited from one inpatient mental health department in a large hospital in China. All participants completed the 12-item version WHODAS 2.0 to assess patients’ functioning in the 30 days prior to the hospitalization. Measurement invariance, including configural, metric and scalar invariance, was tested across patient and proxy groups, using multi-group confirmatory factor analysis. Agreement between patients and proxies was examined by paired Wilcoxon tests and intraclass correlation coefficients (ICC). Subgroup analyses for proxy reliability were conducted within strata of proxy kinship and patient psychiatric diagnosis. Results: The 12-item WHODAS 2.0 achieved configural, metric and partial scalar invariance across patient and proxy groups. Unsatisfactory consistency was found for most items (ICC < 0.75, P < 0.05), especially for items on Cognition, Getting along, Life activities, and Participation in society (ICC < 0.4, P < 0.05). Spouses agreed with patients more often than parents (ICC ≥ 0.4, P < 0.05). The paired Wilcoxon tests found that impairment of patients with psychotic disorders tended to be overestimated by proxies while proxies tended to underestimate impairment of patients with mood disorders. Conclusion: Our study reveals inconsistency between self and proxy reports in the 12-item WHODAS 2.0 among adult patients with mental disorders. When proxy reports is needed, spouses are preferred than parents. We should be aware of proxies’ impairment overestimation among patients with psychotic disorders and underestimation among patients with mood disorders.