It is commonly recognized that a few patients with Graves' disease (GD) develop an overt ophthalmopathy, although most of them show subclinical extraocular muscle enlargement by appropriate imaging techniques. At present, it is not possible to identify the subgroup of GD patients with subclinical retroorbital connective involvement. Recently, it has been shown that increase of soluble intercellular adhesion molecule-1 (sICAM-1) serum levels is correlated to clinical activity score in active Graves' ophthalmopathy (GO) patients with or without hyperthyroidism, suggesting that sICAM-1 serum values could reflect the degree of ocular inflammatory activity. The aim of this longitudinal study was to evaluate sICAM-1 serum levels in GD patients without clinical ophthalmopathy and to assess their possible relationship with occurrence of GO. We measured sICAM-1 serum levels in 103 initially hyperthyroid GD patients without clinical ophthalmopathy and in 100 healthy subjects. All patients were treated with methimazole for 2 yr. Sera were collected from all patients before treatment and then monthly for the first 6 months of therapy, every 2 months in the following 6 months, and finally at the end of the follow-up study. Patients developing GO were excluded from the follow-up at the onset of ophthalmopathy. During the follow-up 17 GD patients (16.5%, group 1) developed overt eye involvement (14 as active inflammatory ophthalmopathy and 3 as ophthalmopathy without clinical retroorbital connective inflammation) and 86 (83.5%, group 2) did not. At start of the study, the mean of sICAM-1 serum concentrations did not differ significantly between the 2 groups, but it was significantly higher than in controls in both groups. No significant correlation between serum sICAM-1 concentrations and free thyroid hormone levels was found in the 2 groups of patients. During the follow-up study, a further increase of sICAM-1 serum levels was observed in 12 of the 14 patients (85.7%) of group 1 who developed active inflammatory ophthalmopathy not only at the onset but also before clinical GO appearance. On the contrary, the 3 patients of group 1 that developed ophthalmopathy without clinical retroorbital inflammation did not show any further increase of sICAM-1 levels at every time of follow-up in comparison with the starting values, even if their sICAM-1 levels were always higher than in normal controls. Finally, group 2 patients showed significantly decreased sICAM-1 levels throughout the follow-up period when compared with the starting values, although they were still significantly higher than in controls. These results indicate that a further increase of sICAM-1 serum levels before the onset of clinical ophthalmopathy may be a marker of subclinical retroorbital connective inflammation in GD patients. Therefore, our study suggests that serial determinations of sICAM-1 serum levels could help to identify and trace at the right time those GD patients prone to developing active inflammatory ophthalmopathy.