Non-Hodgkin lymphoma (NHL) is characterized by a great variability in patient outcomes, resulting in the critical need for identifying new molecular prognostic biomarkers. This study aimed to identify novel circulating prognostic biomarkers based on an miRNA/lncRNA-associated ceRNA network for NHL. Using bioinformatic analysis, we identified the miRNA-lncRNA pairs, and using RT-qPCR, we analyzed their plasma levels in a cohort of 113 NHL patients to assess their prognostic value. Bioinformatic analysis identified SNHG16 and SNHG6 as hsa-miR-20a-5p and hsa-miR-181a-5p sponges, respectively. Plasma levels of hsa-miR-20a-5p/SNHG16 and hsa-miR-181a-5p/SNG6 were significantly associated with more aggressive disease and IPI/FLIPI scores. Moreover, we found that patients with risk expression profiles of hsa-miR-20a-5p/SNHG16 and hsa-miR-181a-5p/SNHG6 presented a higher risk of positive bone marrow involvement. Moreover, hsa-miR-20a-5p/SNHG16 and hsa-miR-181a-5p/SNHG6 pairs' plasma levels were associated with overall survival and progression-free survival of NHL patients, being independent prognostic factors in a multivariate Cox analysis. The prediction models incorporating the ceRNA network expression analysis improved the predictive capacity compared to the model, which only considered the clinicopathological variables. There are still few studies on using the ceRNA network as a potential prognostic biomarker, particularly in NHL, which may permit the implementation of a more personalized management of these patients.