OBJECTIVE: This study aimed to elucidate the network-level intrinsic functional connectivity (iFC) disruptions associated with Parkinson's disease (PD). By focusing on the iFC changes among the areas of the neurodegeneration pattern in the dementia stage of PD (PD-D) and by adopting a mass-univariate iFC analysis approach, the network-based statistics (NBS), we aimed to characterize the network-level dynamics along the steps of cognitive decline in PD with a high statistical sensitivity. METHODS: We conducted surface-based measurements of cortical thickness in PD-D to identify the extent of neurodegeneration, and performed iFC analyses among network nodes localized within this area by utilizing NBS across three cohorts of PD patients: those who are cognitively normal (PD-CN), those with MCI (PD-MCI), and PD-D. An additional interaction analysis was performed to assess how ACE-R scores relate to iFC alterations across cognitive stages. RESULTS: The results demonstrated that declines in iFC start during the PD-MCI stage covering the default mode (DMN), somatomotor (SMN), and dorsal attention networks (DAN), with decreased interactions between the DMN and both the SMN and DAN. As cognitive decline advances to PD-D, iFC decline in limbic and frontoparietal networks are added in addition to reduced interaction of the SMN with the two attention networks, the salience/ventral attention network and DAN. The ACE-R interaction analysis identified cognition-related subnetworks in both PD-MCI and PD-D, with positive associations between ACE-R performance and iFC strength. CONCLUSION: This study demonstrates that iFC alterations differ across cognitive stages of PD in a stage-specific and network-selective manner. While early-stage disruptions primarily involve the DMN and its interactions with attention and motor networks, later stages exhibit a more widespread disintegration. The ACE-R interaction analysis further supports this pattern, showing that lower cognitive performance is systematically associated with reduced connectivity within these same large-scale systems. Particularly striking is the evolving role of the SMN, whose progressive decoupling from attentional and executive systems points to its underestimated contribution to cognitive impairment in PD.