Background: Prognostic factors for diffuse large B-cell lymphoma (DLBCL) in sub-Saharan Africa (SSA) are unknown. We report mature data from one of the first prospective DLBCL cohorts treated under real-world conditions in SSA. Methods: Patients ≥18 years with newly diagnosed DLBCL were enrolled in Malawi from 2013 to 2017. Participants were treated with CHOP chemotherapy, and concurrent antiretroviral therapy (ART) if HIV+. Results: 86 participants were enrolled with mean age 47 (SD 13). 54 (63%) were male, and 51 (59%) were HIV+, of whom 34 (67%) were on ART at DLBCL diagnosis. Median CD4 count was 113 cells/mL (IQR 62-227) and 25 (49%) had an HIV viral load < 400 copies/mL. HIV+ participants were younger and more urban, but otherwise similar to HIV-. 10 (12%) participants died before chemotherapy. Participants received a median 6 CHOP cycles (IQR 4-6). 28% of cycles were delayed or reduced for toxicity, more commonly in the HIV+ group, typically for neutropenia. No patients were lost to follow-up with median follow up 24 months (IQR 16-40) for patients still alive at administrative censoring. 2-yr overall survival (OS) was 38% (95% CI 28-49), and 42% (95% CI 30-53) for patients surviving to CHOP initiation. For those receiving CHOP, 13/43 (30%) deaths were treatment-related, and occurred primarily in patients with adverse baseline DLBCL characteristics. In multivariate analyses, only ECOG performance status (PS) ≥2 and lactate dehydrogenase (LDH) > 2x upper limit of normal (ULN) were associated with mortality. HIV was not associated with OS, but for HIV+ patients, being on ART prior to lymphoma diagnosis was associated with mortality. A simplified prognostic model of LDH > 2x ULN and PS≥2 outperformed the traditional age-adjusted IPI. Conclusions: DLBCL can be successfully treated in SSA and outcomes were not different for HIV+ and HIV- patients, although OS was worse than resource-rich settings. For HIV+ participants, those developing DLBCL on ART had worse OS, suggesting possible effects of the immunologic environment on tumor biology. A simplified prognostic model utilizing only PS and LDH may be optimal in resource-limited settings, where staging is imprecise, but requires additional validation.