Subclavian vein cannulation was suggested as a temporary vascular access for hemodialysis since one of its advantages was considered to be no damage to blood vessels. As we observed six patients with symptomatic subclavian vein thrombosis among 148 patients having received subclavian vein cannulation for hemodialysis, we systematically performed subclavian venogram in 42 asymptomatic patients selected on the basis of a history of previous subclavian vein cannulation. Venograms were performed 15.7 +/- 8.9 months after the removal of the last catheter. Eight patients (19%) had complete thrombosis or severe stenosis of the subclavian vein while six patients (14%) had minimal luminal defects. Considering together the 48 patients, the group with thrombosis or severe stenosis (group 1, n = 14) was compared with the group with minimal defects or normal venograms (group 2, n = 34). In group 1, as compared with group 2, there were more female (64% vs 32%, p = 0.02), more cannulations per vein (1.87 +/- 0.35 vs 1.32 +/- 0.08, p less than 0.05) and more cumulative days of cannulation per vein (35.1 +/- 7.9 vs 24.4 +/- 1.1, p less than 0.001). No difference between the two groups was seen for the number of catheter infections, the number of catheters with poor flow or obstruction, the coagulation screening of the patients or the time-length between the removal of the last catheter and the venogram study. Two of the initially asymptomatic patients developed later on clinical problems related to the subclavian vein thrombosis. We conclude that the subclavian vein cannulation leads to significant damages of the vessels, excluding a whole arm, for future vascular access in some patients.