Modularity plays an important role in brain networks’ architecture and influences its dynamics and the ability to integrate and segregate different modules of cerebral regions. Alterations in community structure are associated with several clinical disorders, specially schizophrenia, although its time evolution is not clear yet. In the present work, we analyze fMRI functional networks of $65$ healthy subjects (HC) and $44$ patients of schizophrenia (SZ), $28$ of them in a chronic state (CR) of illness, and $16$ at early stage (ES). We find clear differences in edges’ weights distribution, networks density, community structure consistency and robustness against edge removal. In comparison to healthy subjects, we found that networks from SZ patients exhibits wider weight distribution, larger overall connectivity, and are more consistent in the community structure across subjects. We also showed that the networks of SZ patients tend to be more robust to edge removal than healthy subjects, while having lower network density. In the case of early stages patients, we found that their networks exhibit topological features consistently in between the ones obtained from the other two groups, resulting in a tendency towards the chronic group state.