Background: Relapsed or refractory diffuse large B-cell lymphoma (DLBCL) is an aggressive cancer with a median overall survival of less than 6 months. We aimed to assess the response to single-agent selinexor, an oral selective inhibitor of nuclear export, in patients with relapsed or refractory DLBCL who had no therapeutic options of potential clinical benefit. Methods: SADAL was a multicentre, multinational, open-label, phase 2b study done in 59 sites in 19 countries. Patients aged 18 years or older with pathologically confirmed diffuse large B-cell lymphoma, an Eastern Cooperative Oncology Group performance status of 2 or less, who had received two to five lines of previous therapies, and progressed after or were not candidates for autologous stem-cell transplantation were enrolled. Germinal centre B-cell or non-germinal centre B-cell tumour subtype and double or triple expressor status were determined by immunohistochemistry and double or triple hit status was determined by cytogenetics. Patients received 60 mg selinexor orally on days 1 and 3 weekly until disease progression or unacceptable toxicity. The study was initially designed to evaluate both 60 mg and 100 mg twice-weekly doses of selinexor; however, the 100 mg dose was discontinued in the protocol (version 7.0) on March 29, 2017, when an improved therapeutic window was observed at 60 mg. Primary outcome was overall response rate. The primary outcome and safety were assessed in all patients who received 60 mg selinexor under protocol version 6.0, or enrolled under protocol versions 7.0 or higher and received at least one dose of selinexor. This trial is registered at ClinicalTrials.gov, NCT02227251 (active but not enrolling). Findings: Between Oct 21, 2015, and Nov 2, 2019, 267 patients were randomly assigned, with 175 allocated to the 60 mg group and 92 to the discontinued 100 mg group. 48 patients assigned to the 60 mg group were excluded due to enrolment before version 6.0 of the protocol; the remaining 127 patients received selinexor 60 mg and were included in analyses of primary outcome and safety. The overall response rate was 28% (36/127; 95% CI 20·7–37·0); 15 (12%) achieved a complete response and 21 (17%) a partial response. The most common grade 3–4 adverse events were thrombocytopenia (n=58), neutropenia (n=31), anaemia (n=28), fatigue (n=14), hyponatraemia (n=10), and nausea (n=8). The most common serious adverse events were pyrexia (n=9), pneumonia (n=6), and sepsis (n=6). There were no deaths judged as related to treatment with selinexor. Interpretation: Single-drug oral selinexor induced durable responses and had a manageable adverse events profile in patients with relapsed or refractory DLBCL who received at least two lines of previous chemoimmunotherapy. Selinexor could be considered a new oral, non-cytotoxic treatment option in this setting. Funding: Karyopharm Therapeutics Inc.