Diffuse large B-cell lymphoma (DLBCL) is the most common B-cell non-Hodgkin lymphoma. To treat this aggressive disease, R-CHOP, a combination of immunotherapy (R; rituximab) and chemotherapy (CHOP; cyclophosphamide, doxorubicin, vincristine, and prednisone), remains the most commonly used regimen for newly diagnosed DLBCLs. However, up to one-third of patients ultimately becomes refractory to initial therapy or relapses after treatment, and the high mortality rate highlights the urgent need for novel therapeutic approaches based upon selective molecular targets. In order to understand the molecular mechanisms underlying relapsed DLBCL, we studied differences in the lipid and metabolic composition of nontreated and R-CHOP-resistant tumors, using a combination of in vivo DLBCL xenograft models and mass spectrometry imaging. Together, these techniques provide information regarding analyte composition and molecular distributions of therapy-resistant and sensitive areas. We found specific lipid and metabolic profiles for R-CHOP-resistant tumors, such as a higher presence of phosphatidylinositol and sphingomyelin fragments. In addition, we investigated intratumor heterogeneity and identified specific lipid markers of viable and necrotic areas. Furthermore, we could monitor metabolic changes and found reduced adenosine triphosphate and increased adenosine monophosphate in the R-CHOP-resistant tumors. This work highlights the power of combining in vivo imaging and MSI to track molecular signatures in DLBCL, which has potential application for other diseases.