To determine the prognostic factors for splenectomy in patients with chronic idiopathic thrombocytopenic purpura (ITP), 26 consecutive patients who had undergone splenectomy as a treatment for ITP in the last 20 years were investigated retrospectively. Predictive values of age, sex, the interval between diagnosis and surgery, presence of antiplatelet antibodies or accessory spleen, bleeding time, response to corticosteroids, response to high-dose immunoglobulin, weight of spleen, serum platelet-associated immunoglobulin G, preoperative maximal platelet count, and preoperative minimal platelet count were examined with multivariate analysis by multiple regression. Age less than 50 years at surgery, platelet count of 100 x 10(9)/l or more in response to high-dose immunoglobulin, and a maximal platelet count of 100 x 10(9)/l or more before splenectomy were favorable prognostic factors (p < 0.05). The correlation coefficient, sensitivity, specificity, and positive-predictive value of the analyzed multiple regression based on prognostic factors were 0.885, 68.8%, 85.7% and 91.7%, respectively. In conclusion, splenectomy is an effective treatment for younger patients, large maximal preoperative platelet counts and preoperative immunoglobulin dependence. Multivariate function analysis was useful for predicting outcome.