High-resolution carotid ultrasonography is considered a fundamental technique for the investigation of the vascular system. However, it is still very unclear whether ultrasonographic studies of carotid arteries are useful for the prediction of cardiovascular events in patients with end-stage renal disease. The prediction power of carotid ultrasonography for all-cause and cardiovascular mortality was tested in a cohort of 138 patients receiving chronic dialysis treatment (91 receiving hemodialysis treatment and 47 receiving continuous ambulatory peritoneal dialysis treatment; follow-up, 29.8 +/- 15.0 mo), and the relationship between this parameter and alterations in left ventricular mass (LVM) and geometry was examined. On univariate analysis, intima media thickness (IMT) was directly related to LVM as well as to the absolute and relative thicknesses of LV walls but independent of LV end-diastolic volume. Data analysis based on LV geometry patterns revealed that patients with concentric hypertrophy were those with the highest IMT. The internal diameter of the common carotid artery (DCCA) was also related to concentric hypertrophy, but the strength of this relationship was of borderline significance (P = 0.06). During the follow-up period, 63 patients died: 32 (51%) of them of cardiovascular causes. IMT was significantly higher (P = 0.006) in patients who died of cardiovascular causes (1.10 +/- 0.21 mm) than in patients who survived (0.99 +/- 0.24 mm), In a Cox regression model, this parameter turned out to be an independent predictor of cardiovascular death, and it retained an independent effect in a model that included LVM. Treatment modality failed to independently predict this outcome. The risk of cardiovascular death was progressively higher from the first IMT tertile onward. DCCA failed to predict cardiovascular outcomes. IMT in dialysis patients is associated with LV concentric hypertrophy and is an independent predictor of cardiovascular death. IMT may be usefully applied for risk stratification in the dialysis population.