Gene expression profiling (GEP), which can divide DLBCL into three groups, is impractical to perform routinely. Although algorithms based on immunohistochemistry (IHC) have been proposed as a surrogate for GEP analysis, the power of them has diminished since rituximab added to the chemotherapy. We assessed the prognostic value of four conventional algorithms and the genes in each and out of algorithm by IHC and fluorescence in situ hybridization in DLBCL patients receiving immunochemotherapy. The results showed that neither single protein within algorithms nor the IHC algorithms themselves had strong prognostic power. Using MYC aberrations (MA) either on the genetic or protein levels, we established a new algorithm called MA that could divide patients into distinct prognostic groups. Patients of MA had much shorter overall survival (OS) and progression-free survival (PFS) than non-MA (2-year OS: 56.9% vs. 98.7%; 2-year PFS: 26.8% vs. 86.9%; P < 0.0001 for both). In conclusions, using additional prognostic markers not associated with cell of origin may accurately predict outcomes of DLBCL. Studies with larger samples should be performed to confirm our algorithm and optimize the prognostic system of DLBCL.