Progression of diabetic retinopathy resists arrest even after institution of intensive glycemic control, suggesting a “metabolic memory” phenomenon, but the mechanism responsible for this phenomenon is still elusive. Gene expression and biological processes can also be regulated by long noncoding RNAs (LncRNAs), the RNAs with >200 nucleotides and no open reading frame for translation, and several LncRNAs are aberrantly expressed in diabetes. Our aim was to identify retinal LncRNAs that fail to reverse after termination of hyperglycemia. Microarray analysis was performed on retinal RNA from streptozotocin-induced diabetic rats in poor glycemic control for 8 months, followed by in good glycemic control (blood glucose >400 mg/dL), or for 4 months, with four additional months of good glycemic control (blood glucose <150 mg/dL). Differentially expressed LncRNAs and mRNAs were identified through Volcano filtering, and their functions were predicted using gene ontology and pathway enrichment analyses. Compared with age-matched normal rats, rats in continuous poor glycemic control had >1479 differentially expressed LncRNAs (710 downregulated, 769 upregulated), and among those, 511 common LncRNAs had similar expression in Diab and Rev groups (139 downregulated, 372 upregulated). Gene Ontology/pathway analysis identified limited LncRNAs in biological processes, but analysis based on biological processes/molecular function revealed >350 genes with similar expression in Diab and Rev groups; these genes were mainly associated with stress response, cell death, mitochondrial damage and cytokine production. Thus, identifying retinal LncRNAs and their gene targets that do not benefit from termination of hyperglycemia have potential to serve as therapeutic targets to slow down the progression of diabetic retinopathy.