We prospectively tested, at diagnosis in essential thrombocythemia (ET) patients with no clear indication to platelet (PLT)-lowering treatment, a scoring system based on age, PLT level, cardiovascular diseases, previous thrombotic events, smoking and dysmetabolic diseases. From 04/92 to 03/98, 168 consecutive patients were enrolled. Hydroxyurea (HU) was started at diagnosis in 32 "symptomatic" patients and in 33 patients aged >70 years. The remaining 103 patients ("asymptomatic" and aged <70 years) were classified according to our scoring system. Thirty-two patients with score > or = 4 started HU early after diagnosis. The remaining 71 patients with score <4 at diagnosis received anti-aggregating agents only; of them, 24 (33.8%) started HU during follow-up after a median time from diagnosis of 28 months, while 47 (66.2%) did not start any PLT-lowering treatment. Thrombotic complications occurred in 9/103 patients (8.7%); in particular, they occurred in 4/32 patients (12.5%) with score > or = 4 receiving HU since diagnosis and in 5/71 (7%) with score <4 under anti-aggregating agents only. This scoring system appears effective to discriminate a different risk of thrombotic events, and could be useful to decide when a PLT-lowering therapy needs to be started.