Atherosclerosis is the major cause of death in patients with diabetes mellitus, accounting for more than 70 % of mortality in all forms of the disease [1,2]. There is plenty of evidence that monocytes are involved in the pathogenesis of atherosclerosis [3,4], the earliest morphological evidence of the disease being the binding of monocytes to the endothelium [5][6][7]. This adhesion of monocytes, a prerequisite for their recruitment and accumulation in the lesion, is probably due to the appearance of adhesion Diabetologia (1997) Summary Although elevated levels of soluble E-selectin and intercellular cell adhesion molecules-1 (ICAM-1) have been reported in non-insulin-dependent diabetes mellitus (NIDDM), it is not clear by what mechanism this elevation occurs and whether or not it is related to glycaemic control. In this study we analyse: 1) the relation of glycaemic control with the concentrations of E-selectin, vascular cell adhesion molecules-1 (VCAM-1) and ICAM-1 in NIDDM patients; 2) whether metabolic control can affect the oxidative stress (as measured by plasma hydroperoxide concentration and susceptibility of LDL to in vitro oxidation) and hence the adhesion molecule plasma concentrations. Thirty-four (19 males and 15 females) poorly controlled NIDDM patients were studied. All parameters were evaluated at the beginning of the study and after 90 days of dietary and pharmacological treatment. The treatment decreased HbA 1C (p < 0.001), E-selectin (p < 0.001), plasma hydroperoxides (p < 0.003) and the susceptibility of LDL to in vitro oxidation (lag phase) (p < 0.0001). Before treatment HbA 1C , lag phase and lipid hydroperoxides correlated with E-selectin plasma concentration (r = 0.51, -0.57 and 0.54, respectively, p < 0.01). There was also a correlation between HbA 1C and lag phase (p < 0.01) and between HbA 1C and lipid hydroperoxides (p < 0.01). In addition, the variations of HbA 1C , lag phase and lipid hydroperoxide values correlated with those for E-selectin concentration after 90 days' treatment (r = 0.54, -0.64 and 0.61, respectively, p < 0.01). In multiple linear correlation analysis, however, the partial correlation coefficients of HbA 1C (basal and variations) with Eselectin concentration (basal and variations) fell to non-significant values (r = 0.12 and 0.25, respectively) when LDL lag phase and plasma hydroperoxides were kept constant. The results indicate that the improvement of metabolic control in NIDDM patients is associated with a decrease of E-selectin plasma levels; they also suggest that glycaemic control per se is not directly implicated in determining E-selectin plasma concentration; glycaemic control could affect E-selectin concentration through its effect on oxidative stress. [Diabetologia (1997) 40: 584-589]