Purpose To develop and validate a risk score for relapse in the CNS in patients with diffuse large B-cell lymphoma (DLBCL). Patients and Methods A total of 2,164 patients (18 to 80 years old) with aggressive B-cell lymphomas (80% DLBCL) treated with rituximab and CHOP (cyclophosphamide, doxorubicin, vincristine, prednisone)-like chemotherapy, who were enrolled in studies from the German High-Grade Non-Hodgkin Lymphoma Study Group and the MabThera International Trial, were analyzed for occurrence of relapse/progression in the CNS. The resulting risk model was validated in an independent data set of 1,597 patients with DLBCL identified in the British Columbia Cancer Agency Lymphoid Cancer database. Results The risk model consists of the International Prognostic Index (IPI) factors in addition to involvement of kidneys and/or adrenal glands (CNS-IPI). In a three-risk group model, the low-risk group (46% of all patients analyzed), the intermediate-risk group (41%), and the high-risk group (12%) showed 2-year rates of CNS disease of 0.6% (CI, 0% to 1.2%), 3.4% (CI, 2.2% to 4.4%), and 10.2% (CI, 6.3% to 14.1%), respectively. Patients from the validation British Columbia Cancer Agency data set showed similar rates of CNS disease for low-risk (0.8%; CI, 0.0% to 1.6%), intermediate-risk (3.9%; CI, 2.3% to 5.5%), and high-risk (12.0%; CI, 7.9% to 16.1%) groups. Conclusion The CNS-IPI is a robust, highly reproducible tool that can be used to estimate the risk of CNS relapse/progression in patients with DLBCL treated with R-CHOP (rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone) chemotherapy. Close to 90% of patients with DLBCL belong to the low- and intermediate-risk groups and have a CNS relapse risk < 5%; they may be spared any diagnostic and therapeutic intervention. In contrast, those in the high-risk group have a > 10% risk of CNS relapse and should be considered for CNS-directed investigations and prophylactic interventions.