The introduction of rituximab has improved the overall prognosis of diffuse large B-cell lymphoma (DLBCL). However, the impact of rituximab on central nervous system (CNS) involvement in DLBCL remains a matter of debate. Patients with DLBCL and no CNS involvement at initial diagnosis were eligible for this analysis. Patients must have received treatment either with CHOP (cyclophosphamide, doxorubicin, vincristine and prednisone) or CHOP plus rituximab (R-CHOP). We analyzed the incidence, clinical features and outcomes of CNS involvement that developed during or after completion of therapy. A cohort of 599 patients was eligible for this analysis. With a median follow-up of 26 and 21 months, respectively, 19 of 294 (6.5 %) in the CHOP group and 13 of 305 (4.3 %) in the R-CHOP group developed CNS involvement. Rituximab did not significantly reduce the risk of CNS involvement either in the univariate (P = 0.354) or in the multivariate analysis (RR 0.632, 95 % CI 0.301–1.327, P = 0.225). No patient developed CNS disease after 19 months in the R-CHOP group whereas four patients (21.1 %) in the CHOP group developed CNS disease 2 years after initial diagnosis (range 34–83 months). Systemic disease prior to or coincident with CNS occurrence was more common in the CHOP group than in the R-CHOP group (73.7 versus 38.5 %, P = 0.046). Isolated CNS events were more common in the R-CHOP group than those in the CHOP group (53.8 versus 10.5 %, P = 0.015). This study indicates that isolated CNS events are more common in DLBCL patients treated with R-CHOP than those treated with CHOP alone. Our data also suggest that the time and pattern of CNS events and systemic disease status differ with the addition of rituximab. Better methods for earlier detection and prophylaxis of CNS involvement are needed in the rituximab era.