The dosage of soluble programmed cell death ligand 1 (sPD-L1) protein in the blood of adults with cancer has never been performed in a prospective patient cohort. We evaluated the clinical impact of sPD-L1 level measured at the time of diagnosis for newly diagnosed diffuse large B-cell lymphoma (DLBCL). Soluble PD-L1 was measured in the plasma of 288 patients enrolled in a multicenter, randomized phase III trial that compared R-high-dose chemotherapy with R-CHOP. The median follow-up was 41.4 months. A cutoff of 1.52 ng/ml of PD-L1 level was determined and related to overall survival (OS). Patients with elevated sPD-L1 experienced a poorer prognosis with a 3-year OS of 76% versus 89% (P<0.001). Considering clinical characteristics, the multivariate analysis retained this biomarker besides bone marrow involvement and abnormal lymphocyte-monocyte score as independently related to poor outcome. sPD-L1 was detectable in the plasma and not in the serum, found elevated in patients at diagnosis compared with healthy subjects and its level dropped back to normal value after CR. The intention-to-treat analysis showed that elevated sPD-L1 was associated with a poorer prognosis for patients randomized within the R-CHOP arm (P<0.001). Plasma PD-L1 protein is a potent predicting biomarker in DLBCL and may indicate usefulness of alternative therapeutic strategies using PD-1 axis inhibitors.