AbstractHigh‐dose intravenous methotrexate (HD‐MTX) CNS prophylaxis in high‐risk diffuse large B cell lymphoma (DLBCL) remains controversial. We describe real‐world CNS relapse incidence following baseline cerebrospinal fluid (CSF) analysis to exclude asymptomatic leptomeningeal involvement in newly diagnosed high‐risk DLBCL patients with versus without single‐route HD‐MTX CNS prophylaxis. Consecutively diagnosed high‐risk systemic DLBCL patients without leptomeningeal involvement by CSF analysis (noCNS) were identified retrospectively. Five‐year CNS relapse incidence and survival outcomes were examined, as stratified by receipt of HD‐MTX prophylaxis. Secondary analysis of survival outcomes in patients with synchronous leptomeningeal involvement (CNSinv) by CSF analysis at diagnosis were compared with the noCNS group. No significant difference in 5‐year CNS relapse incidence was observed following HD‐MTX prophylaxis versus no prophylaxis (total n = 445) despite similar CNS‐IPI risk; 6.2% versus 5.6%, adjusted HR 1.08 (95% CI 0.41–2.85), p = .88; nor in 5‐year progression free survival (PFS) or overall survival (OS) risk. Of CNSinv patients, 93.3% had ≥1 extranodal site. Increased CNS relapse/progression risk (5‐year risk; HR 10.7 [95% CI 5.35–21.37], p < .0001) and inferior PFS and OS were observed in CNSinv versus all noCNS patients. The CNSinv group had superior OS compared with noCNS patients who later experienced CNS relapse (HR 0.55, p = .052). HD‐MTX prophylaxis does not reduce CNS relapse risk in high‐risk systemic DLBCL without leptomeningeal involvement by CSF analysis at diagnosis. Asymptomatic patients with synchronous leptomeningeal involvement on baseline CSF examination are at increased risk of further CNS disease events and inferior survival compared to patients without CSF involvement.