Cancer of unknown primary (CUP) has a poor prognosis. Given the recent approval of immune checkpoint inhibitors for several cancer types, we carried out a multicenter phase II study to assess the efficacy of nivolumab for patients with CUP. Patients with CUP who were previously treated with at least one line of systemic chemotherapy constituted the principal study population. Previously untreated patients with CUP were also enrolled for exploratory analysis. Nivolumab (240 mg/body) was administered every 2 weeks for up to 52 cycles. The primary endpoint was objective response rate in previously treated patients as determined by blinded independent central review according to RECIST version 1.1. Fifty-six patients with CUP were enrolled in the trial. For the 45 previously treated patients, objective response rate was 22.2% 95% confidence interval (CI), 11.2% to 37.1%, with a median progression-free survival and overall survival of 4.0 months (95% CI, 1.9-5.8 months) and 15.9 months (95% CI, 8.4-21.5 months), respectively. Similar clinical benefits were also observed in the 11 previously untreated patients. Better clinical efficacy of nivolumab was apparent for tumors with a higher programmed death-ligand 1 expression level, for those with a higher tumor mutation burden, and for microsatellite instability-high tumors. In contrast, no differences in efficacy were apparent between tumor subgroups based on estimated tissue of origin. Adverse events were consistent with the known safety profile of nivolumab. No treatment-related death was observed. Our results demonstrate a clinical benefit of nivolumab for patients with CUP, suggesting that nivolumab is a potential additional therapeutic option for CUP. •Nivolumab demonstrated clinical benefits in a statistically assessable number of patients with CUP.•Objective response rate was 22.2% in previously treated patients with CUP, which met the study primary endpoint.•Benefits were more apparent in patients with known biomarkers for ICIs, but also observed in those without such markers.•No differences in efficacy were apparent between tumor subgroups based on estimated tissue of origin.•Nivolumab could be an additional therapeutic option for these unmet medical needs.