The Nordic idiopathic thrombocytopenic purpura study data showed that morbidity occurred mainly in children with thrombocytopenia lasting >3 months, whereas, the risk period with platelet counts <20 x 10(9)/l was short and the number of bleeding events low in children with shorter disease duration. These brief, uneventful courses were predicted by developing a scoring system based on six clinical features: abrupt onset (weight 5), age <10 years (3), preceding infection (2), platelet count <5 x 10(9)/l, wet purpura (1) and male gender (1). The score was derived and validated in two different cohorts of children. High scores (10-14) clearly identified low-risk patients. The score provides valid prognostic information and may be useful in clinical decision-making.