Negative symptoms in treatment-resistant schizophrenia (TRS) are notably persistent and minimally affected by antipsychotics, the transcutaneous auricular vagus nerve stimulation (taVNS) is a promising treatment approach. However, clinical trials are scarce, and further efficacy data are needed. We conducted a double-blind, sham-controlled, randomized clinical trial to determine the efficacy and safety of taVNS as an add-on treatment for patients with TRS with predominantly negative symptoms and to investigate potential biomarkers of efficacy. A total of 50 patients underwent a two-week intervention of active taVNS (n = 25) or sham taVNS (n = 25), followed by a two-week follow-up. Primary outcome was the change in the PANSS-factor score for negative symptoms (PANSS-FSNS) assessed after the intervention. In the intention-to-treat analysis, patients receiving active taVNS showed a significantly greater improvement in negative symptoms compared with those receiving the sham procedure (PANSS-FSNS difference, -1.36; effect size, -0.62; 95% CI, -1.20 to -0.04; p = 0.033), with effects sustained at follow-up and good tolerability. Inflammatory cytokines and EEG coherence showed that in the active group, the change in PANSS-FSNS scores after treatment was significantly correlated with changes in tumour necrosis factor (TNF)-α (r = 0.56, corrected p = 0.017) and beta-band coherence between the left frontal and parietal regions (r = -0.56, p = 0.004), but not in the sham group. This study suggests that taVNS may effectively and safely ameliorate negative symptoms in TRS, with TNF-α and beta-band coherence between the left frontal and parietal regions as potential sensitivity efficacy biomarkers. Chinese Clinical Trial Registry ( http://www.chictr.org.cn .), ChiCTR2400085198.