Contractile dysfunction is underdiagnosed in early stages of diabetic cardiomyopathy. We evaluated the potential of circulating long non-coding RNAs (lncRNAs) as biomarkers of subclinical cardiac abnormalities in type 2 diabetes. Forty-eight men with well-controlled type 2 diabetes and 12 healthy age-matched volunteers were enrolled in the study. Left ventricular (LV) parameters were measured by magnetic resonance imaging. A panel of lncRNAs was quantified in serum by RT-qPCR. No differences in expression levels of lncRNAs were observed between type 2 diabetes patients and healthy volunteers. In patients with type 2 diabetes, long intergenic non-coding RNA predicting cardiac remodeling (LIPCAR) was inversely associated with diastolic function, measured as E/A peak flow (P < 0.050 for all linear models). LIPCAR was positively associated with grade I diastolic dysfunction (P < 0.050 for all logistic models). Myocardial infarction-associated transcript (MIAT) and smooth muscle and endothelial cell-enriched migration/differentiation-associated long noncoding RNA (SENCR) were directly associated with LV mass to LV end-diastolic volume ratio, a marker of cardiac remodelling (P < 0.050 for all linear models). These findings were validated in a sample of 30 patients with well-controlled type 2 diabetes. LncRNAs are independent predictors of diastolic function and remodelling in patients with type 2 diabetes.