Aims. Diffuse large B-cell lymphomas (DLBCL) can be divided into different subgroups (germinal center B-cell-like [GCB] and non-GCB) according to their gene expression profiles. Immunohistochemistry has been proposed as a surrogate for identifying these subgroups, but data about its efficacy in providing prognostic information are conflicting. Methods and results. This study retrospectively analyzed a series of 105 DLBCL, defined as GCB and non-GCB according to CD10, bcl-6, and MUM1 expression. All patients received a first-line anthracycline-based (CHOP-like) chemotherapy. A total of 50 patients (48%) were identified as GCB and 55 (52%) as non-GCB. The overall response rate was 89% (94/105), with 62 (59%) complete response. Disease progressions were equally distributed between the 2 subgroups and were not significantly different (P = .756) considering the primary site of involvement (nodal or extranodal). The median follow-up was 62 months (range 5-126 months). Overall survival at 5 years was not significantly different between the groups (P = .3468) and was 72.3% and 66.6% for GCB and non-GCB, respectively. Conclusion. The results do not support the prognostic value of GCB and non-GCB immunohistochemical categories in DLBCL of both nodal and extranodal origin. Furthermore, a limited number of antigens may be not sufficient to identify the same patterns defined by cDNA microarray.