Mild traumatic brain injury (mTBI) is the most prevalent neurological insult and leads to long-lasting cognitive impairment. Neuroimaging studies have discovered abnormalities in brain network connectivity following mTBI as the underlying neural basis of cognitive deficits. However, the pathophysiologic mechanisms involved in imaging alterations remain elusive. Proteins neuron-specific enolase (NSE) and ubiquitin C terminal hydrolase 1 (UCHL1) are reliable markers for neuronal cell-body damage, both of which have been demonstrated to be increased in serum following mTBI. Therefore, we conducted a longitudinal study to examine relationships between abnormal brain network connectivity and serum neuronal biomarkers and their associations with cognitive recovery following mTBI. Sixty patients were followed-up at 1week and 3 months post-injury and 41 controls were recruited. Resting-state functional MRI was used to build a functional connectivity matrix within large-scale intrinsic networks, their topological properties were analysed using graph theory measures. We found that, compared to controls, mTBI patients showed significant decreases in a number of nodal characteristics in default mode network (DMN), salience network (SN) and executive network (EN) (P < 0.05, FDR corrected) at 3 months post-injury. Linear regression analysis found elevated serum NSE in acute phase could predict lower efficiency and degree centrality of anterior DMN at 3 months post-injury. And efficiency and degree centrality of anterior DMN were negatively associated with working memory. Our study showed neuronal injury was associated with alterations in brain network connectivity after mTBI. These findings can facilitate capability to predict the brain functional outcomes and cognitive recovery in mTBI.