Cerebral cortex encodes crucial information of brain development, cytoarchitecture and function. However, varying data acquisition conditions at different centers could hamper group-wisely statistical analysis. This study aims to test the consistency of cortical thickness in the human brain across four sites and harmonize the deviations. Our results showed that variation of cortical thickness across sites were regionally independent, and deviation across centers could be reduced by linear regression method at a global scale, while the variations across subjects were well preserved. Those results suggest that our method has the promise in harmonizing cortical thickness measures in multi-center study.