Addition of immune checkpoint inhibitors to neoadjuvant chemotherapy (NACT) is a promising strategy in early breast cancer, but the optimal duration of therapy is currently unknown. In the GeparNuevo (NCT02685059) trial, addition of durvalumab to NACT as previously reported led to a moderate increase in pathological complete response (pCR) rate by an absolute 9% (P = 0.287). Patients with cT1b-cT4a-d triple-negative breast cancer (TNBC) received durvalumab 1.5 g or placebo every 4 weeks added to nab-paclitaxel 125 mg/m2 weekly for 12 weeks, followed by durvalumab/placebo every 4 weeks plus epirubicin/cyclophosphamide every 2 weeks followed by surgery. Durvalumab was not continued after surgery. The primary objective was pCR. Secondary endpoints included invasive disease-free survival (iDFS), distant disease-free survival (DDFS) and overall survival (OS). A total of 174 patients were randomised between June 2016 and October 2017. After a median follow-up of 43.7 months, 34 events had occurred. Despite a non-significant increase in the pCR rate, significant differences were observed for 3-year iDFS, DDFS and OS: iDFS was 85.6% with durvalumab versus 77.2% with placebo hazard ratio (HR) 0.48, 95% confidence interval (CI) 0.24-0.97, stratified log-rank P = 0.036; DDFS 91.7% versus 78.4% (HR 0.31, 95% CI 0.13-0.74, P = 0.005); OS 95.2% versus 83.5% (HR 0.24, 95% CI 0.08-0.72, P = 0.006). pCR patients had 3-year iDFS of 95.5% with durvalumab and 86.1% without (HR 0.22, 95% CI 0.05-1.06). In the non-pCR cohort 3-year iDFS was 76.3% versus 69.7% (HR 0.67, 95% CI 0.29-1.54). Multivariable analysis confirmed a durvalumab effect independent of the pCR effect. No new safety signals occurred. Durvalumab added to NACT in TNBC significantly improved survival despite a modest pCR increase and no adjuvant component of durvalumab. Additional studies are needed to clarify the optimal duration and sequence of checkpoint inhibitors in the treatment of early TNBC. •Significant improvement of secondary endpoints iDFS, DDFS and OS by adding durvalumab to neoadjuvant chemotherapy.•Survival improvement although no adjuvant checkpoint inhibitor therapy was given.•Long-term effect is seen in pCR as well as non-pCR patients.•The improved survival outcome with checkpoint inhibitor therapy is only partially explained by the increased pCR rate.•Our results suggest that additional long-term antitumour effects are present.