Introduction. This study examined the influence of diabetes on left ventricular hypertrophy (LVH) detected by different electrocardiographic (ECG) criteria and its changes induced by a 12-month candesartan-based regimen. Methods. The patients were treated for a 12-month period with a candesartan-based regimen (8/16 mg + hydrochlorothiazide12.5 mg + additional drugs to target blood pressure < 140/90 mmHg [< 130/80 in diabetics]). Cornell voltage index (CorV), Cornell product (CorP), Sokolow-Lyon voltage index (SokV), and Sokolow-Lyon product (SokP) were calculated. In total 276 patients were included, 51 with diabetes. Results. At study end, blood pressure was reduced 19.0±9.2/7.3±3.4 mmHg in diabetic patients and 18.8±9.1/8.0±3.2 mmHg in non-diabetic subjects (both p<0.01 vs. baseline; p=0.85 between groups).At baseline, 37.5% of diabetic and 26.4% of non-diabetic patients fulfilled criteria of ECG-LVH by CorP (p=0.02), 25.7% and 23.2%, respectively, by SokP (p=0.18), 11.8% and 13.7% by CorV (p=0.16), and 14.3% and 11.6% by SokV (p=0.10).At study end, the prevalence of ECG-LVH was reduced to 25.1% (relative risk reduction [RRR] 33.3%, p=0.001) and 18.2% (RRR 29.2%, p=0.001) by CorP and SokP, in diabetic patients, respectively. In non-diabetic patients, only the CorP criterion showed a significant decrease (RRR 18.9%, p=0.01). No significant changes were observed by other criteria.The RRR of ECG-LVH with treatment was significantly higher in diabetics according to CorP and SokP criteria. Conclusions. The prevalence of ECG-LVH detected by CorP was higher in diabetics. Diabetics achieved higher reductions in ECG-LVH prevalence than non-diabetics with a candesartan-based regimen.