© 2015, Chinese Medical Association. All rights reserved. Background: Optimizing treatment outcomes for depression requires understanding of how evidence-based treatments are utilized in clinical practice. Antipsychotic medications concurrent with antidepressant treatment are frequently used in major depression, but few studies have investigated trends and patterns of their use over time. This study aimed to examine the prescription patterns of antipsychotic medications for major depression in China from 2002 to 2012 and their association with treatment satisfaction and quality of life (QOL). Methods: A total of 3655 subjects with major depression treated in 45 Chinese psychiatric hospitals/centers nationwide were interviewed between 2002 and 2012. Patients’ socio-demographic and clinical characteristics including psychopathology, medication side effects, satisfaction with treatment and QOL were recorded using a standardized protocol and data collection. Results: The frequency of antipsychotic use was 24.9% in the whole sample; the corresponding figures were 17.1%, 20.3%, and 32.8% in 2002, 2006, and 2012, respectively (χ2 = 90.3, df = 2, P <0.001). Multiple logistic regression analyses revealed that patients on concurrent antipsychotics had significantly more delusions or hallucinations, longer illness duration, greater side effects, and more likely to be treated as inpatients and in major hospitals (i.e., Level-III hospital). Antipsychotic use was associated with lower treatment satisfaction while there was no significant difference with respect to physical and mental QOL between the antipsychotic and nonantipsychotic groups. Conclusions: Concurrent antipsychotic use was found in about one in four treated depressed patients in China, which has increased over a 10-year period. Considering the association of drug-induced side effects and the lack of patients’ and relatives’ satisfaction with antipsychotic treatment, further examination of the rationale and appropriateness of the use of antipsychotics in depression is needed.