The aim of this study was to explore the potential of default mode network (DMN) functional connectivity for predicting the success of smoking cessation in patients with tobacco dependence in the context of a real-time function al MRI (RT-fMRI) neurofeedback (NF) supported therapy.Fifty-four tobacco-dependent patients underwent three RT-fMRI-NF sessions including resting-state functional connectivity (RSFC) runs over a period of 4 weeks during professionally assisted smoking cessation. Patients were randomized into two groups that performed either active NF of an addiction-related brain region or sham NF. After preprocessing, the RSFC baseline data were statistically evaluated using seed-based ROI (SBA) approaches taking into account the smoking status of patients after 3 months (abstinence/relapse).The results of the real study group showed a widespread functional connectivity in the relapse subgroup (n?=?10) exceeding the DMN template and mainly low correlations and anticorrelations in the within-seed analysis. In contrast, the connectivity pattern of the abstinence subgroup (n?=?8) primarily contained the core DMN in the seed-to-whole-brain analysis and a left lateralized correlation pattern in the within-seed analysis. Calculated Multi-Subject Dictionary Learning (MSDL) matrices showed anticorrelations between DMN regions and salience regions in the abstinence group. Concerning the sham group, results of the relapse subgroup (n?=?4) and the abstinence subgroup (n?=?6) showed similar trends only in the within-seed analysis.In the setting of a RT-fMRI-NF-assisted therapy, a widespread intrinsic DMN connectivity and a low negative coupling between the DMN and the salience network (SN) in patients with tobacco dependency during early withdrawal may be useful as an early indicator of later therapy nonresponse.