Serum albumin (SA) has been shown to be a prognostic marker in many hematological malignancies and in diffuse large B-cell lymphoma (DLBCL) prior to chemo-immunotherapy. SA may be a surrogate for age, comorbid status, and disease severity. Here, we aimed to assess whether SA can be an independent prognostic marker in patients with newly diagnosed DLBCL treated with rituximab, cyclophosphamide, vincristine, doxorubicin, and prednisone (R-CHOP). Patients who presented at the Moffitt Cancer Center from 2007 to 2010 for DLBCL diagnosis or treatment were identified using our institutional database. Clinical and treatment data were recorded, including SA levels at diagnosis. Survival time was estimated using the Kaplan-Meier method, with Cox proportional hazard model used to identify potential risk factors for time-to-event data. From 295 identified patients, 171 were excluded for not having primary treatment at our institution or not having R-CHOP treatment. In 124 included patients (mean age at diagnosis of 58 years, 91 % Caucasian), 25 % were categorized as poor by the revised International Prognostic Index. Overall and progression-free survival at 4 years were 65 % (95 % CI 57-75) and 58 % (95 % CI 0.49-0.69), respectively. Using multivariate analysis, we found that the hazard index of death of patients with SA ≥3.7 g/dL was 26 % (95 % CI 13-53) of the hazard for those patients who had SA <3.7 g/dL when controlling for the revised International Prognostic Index risk and initial lymphocyte count. Our study shows that SA ≥3.7 g/dL is an independent prognostic marker in DLBCL patients treated with R-CHOP.