In this study we analyzed the prognostic value of single and combined immunohistochemical markers, according to algorithms proposed by Hans et al. and Muris et al. in 66 de novo diffuse large B-cell lymphoma (DLBCL) cases. The main aim of our study was to compare usefulness of these two immunohistochemical algorithms for the subdivision of DLBCL into prognostically relevant subgroups. Cases classified as germinal centre B-cell (GCB) had a significantly lower risk of death (p = 0.008) compared with the non-GCB group. The 5-year overall survival (OS) rate was 85% for the GCB group and only 30% for the non-GCB group (p = 0.003). Furthermore, division into the GCB and non-GCB group predicted prognosis in cases with low International Prognostic Index (IPI) (p = 0.03). GCB patients with a low IPI score had a significantly better OS than those from the non-GCB group (93% versus 45%) (p = 0.02). Although the 5-year OS of favourable group 1 from Muris algorithm was slightly better than in group 2, the difference was not significant (p = 0.241). In summary, our results indicate that the algorithm of Hans et al. has a significantly better prognostic value. By using immunohistochemistry and this algorithm, we can subclassify DLBCL into prognostically distinct subgroups and further refine the prognosis based on IPI.