Objective <p>To determine the prognostic value of metabolic heterogeneity parameter coefficient of variation (COV) measured on baseline ¹⁸F-FDG positron emission tomography/computed tomography (PET/CT) in a series of newly diagnosed patients with diffuse large B-cell lymphoma (DLBCL) treated with the classical R-CHOP or R-CHOP-like chemotherapy.</p> Methods <p>One hundred and three patients with histopathologically proven DLBCL, who underwent ¹⁸F-FDG PET/CT and had available follow-up results were retrospectively enrolled. The clinical data and baseline metabolic parameters of ¹⁸F-FDG PET/CT, including maximum standardized uptake value (SUVmax), tumor metabolic tumor volume (TMTV), bone marrow-to-liver ratio (BLR), COV, and ¹⁸F-FDG uptake in bone marrow (BM) involvement were collected and documented. Progression-free survival (PFS) and overall survival (OS) served as endpoints. The prognostic value of clinical data and metabolic parameters for PFS and OS was evaluated using Kaplan-Meier survival analysis. Based on the multivariate Cox regression analysis, two predictive models for PFS and OS were developed and their predictive performance was assessed.</p> Results <p>At a median follow-up time of 39.6 months (95%CI, 27.55–51.65), 40 patients (38.8%) experienced disease progression and 15 patients (14.6%) died. Patients with high COV had a shorter PFS (median PFS: 7.4 months vs. ‘not reached’, <i>P</i> &lt; 0.001) and OS (no patients died in low COV group, <i>P</i> = 0.001). Based on the independent risk factors obtained from the multivariate Cox regression analyses for PFS (pathological BM involvement, TMTV and COV) and OS (pathological BM involvement and COV), two predictive models were constructed and visualized using nomogram. The calibration analysis and the decision curves demonstrated good performance of the models. ¹⁸F-FDG uptake of BM involvement was also a significant factor influencing PFS (<i>P</i> = 0.036) and OS (<i>P</i> = 0.019). Furthermore, the combination of COV and ¹⁸F-FDG uptake in BM involvement provided significant prognostic stratification for both PFS (<i>P</i> &lt; 0.001) and OS (<i>P</i> = 0.001), successfully categorizing patients into three distinct risk groups.</p> Conclusions <p>Metabolic heterogeneity parameter COV was a strong independent prognostic factor in DLBCL patients. Based on pathological BM involvement, TMTV and COV, two predictive models were established and their performance was evaluated for predicting PFS and OS. The risk stratification combining COV and ¹⁸F-FDG uptake in BM involvement also had significant prognosis value in PFS and OS. Our exploratory study revealed that metabolic heterogeneity parameters measured via ¹⁸F-FDG PET/CT can preliminarily distinguish DLBCL patients most responsive to standard R-CHOP or analogous chemotherapy regimens, providing a preliminary basis for prognostic stratification in this cohort.</p>