Combining immune-checkpoint inhibitors with chemotherapy yielded an increased response rates in patients with metastatic triple-negative breast cancer (TNBC). Therefore, we evaluated the addition of durvalumab to standard neoadjuvant chemotherapy (NACT) in primary TNBC. GeparNuevo is a randomised phase II double-blind placebo-controlled study randomising patients with TNBC to durvalumab or placebo given every 4weeks in addition to nab-paclitaxel followed by standard EC. In the window-phase durvalumab/placebo alone was given 2 weeks before start of nab-paclitaxel. Randomisation was stratified by stromal tumour-infiltrating lymphocyte (sTILs). Patients with primary cT1b-cT4a-d disease, centrally confirmed TNBC and sTILs were included. Primary objective was pathological complete response (pCR) (ypT0 ypN0). A total of 174 patients were randomised, 117 participated in the window-phase. Median age was 49.5years (range 23–76); 47 patients (27%) were younger than 40years; 113 (65%) had stage ≥IIA disease, 25 (14%) high sTILs, 138 of 158 (87%) were PD-L1-positive. pCR rate with durvalumab was 53.4% (95% CI 42.5% to 61.4%) versus placebo 44.2% (95% CI 33.5% to 55.3%; unadjusted continuity corrected χ2P=0.287), corresponding to OR=1.45 (95% CI 0.80–2.63, unadjusted Wald P=0.224). Durvalumab effect was seen only in the window cohort (pCR 61.0% versus 41.4%, OR=2.22, 95% CI 1.06–4.64, P=0.035; interaction P=0.048). In both arms, significantly increased pCR (P<0.01) were observed with higher sTILs. There was a trend for increased pCR rates in PD-L1-positive tumours, which was significant for PD-L1-tumour cell in durvalumab (P=0.045) and for PD-L1-immune cell in placebo arm (P=0.040). The most common immune-related adverse events were thyroid dysfunction any grade in 47%. Our results suggest that the addition of durvalumab to anthracycline-/taxane-based NACT increases pCR rate particularly in patients treated with durvalumab alone before start of chemotherapy. ClinicalTrials.gov number: NCT02685059.