Mild Cognitive Impairment (MCI) is a neurological diagnosis given to people who have impaired cognitive abilities beyond the standard of their age, but that do not interfere significantly with their daily lives. Patients with MCI have a high risk to suffer from Alzheimer’s disease (AD) within a few years. Although AD is a degenerative disease which starts from a local lesion, its effects on cognitive abilities are immediate and global, because AD is generally accepted as a disconnection syndrome. Since human brain is organized into functionally segregated units as well as anatomically distinct parts, we investigated how topological organizations change in brain functional networks of MCI patients. In order to clarify this change, we adopted a community detection algorithm and modular structure analysis from the graph theory. As a result, we found overall community structure was less distinct and less segregated in MCI. Also, the structural stability was decreased in that the bond between brain regions in the same module was weakened. We suggest that this distortion and instability of community structures of functional networks are responsible for cognitive deficits of MCI patients and hope the module analysis might play a role in a diagnosis of MCI and the early stage of AD in addition to traditional clinical diagnosis and structural imaging.