Preference relation has been one of the most useful tools for experts to express their comparison information over alternatives in group decision-making (GDM) problems. Recently, a new type of preference relations called linguistic preference relations with self-confidence (LPRs-SC) has been proposed, which makes multiple self-confidence levels into consideration when experts provide their preferences. This study focuses on the consensus reaching process for GDM with LPRs-SC. To do that, some new operational laws for LPRs-SC are presented. Subsequently, an iteration-based consensus proposal for LPRs-SC is proposed. In the proposal, we aggregate the individual LPRs-SC using a self-confidence indices-based method which gives more importance to the most self-confident experts. A self-confidence score function is presented to derive the individual and collective priority vectors. Moreover, considering experts’ acceptable adjustment range of preference values, a two-step feedback adjustment mechanism is utilized to improve the consensus level, which adjusts both the preference values and the self-confidence levels. Finally, an example and some analyses are furnished to demonstrate the feasibility and effectiveness of the proposed method.