Chronic stress is closely linked to clinical depression, which could be assessed by a chronic unpredictable mild stress (CUMS) animal model. We present here a GC/MS-based metabolic profiling approach to investigate neurochemical changes in the cerebral cortex, hippocampus, thalamus, and remaining brain tissues. Multi-criteria assessment for multivariate statistics could identify differential metabolites between the CUMS-model rats versus the healthy controls. This study demonstrates that the significantly perturbed metabolites mainly involving amino acids play an indispensable role in regulating neural activity in the brain. Therefore, results obtained from such metabolic profiling strategy potentially provide a unique perspective on molecular mechanisms of chronic stress.