The Vogt-Koyanagi-Harada (VKH) syndrome is a form of recurrent uveitis with often a poor long-term visual acuity. The risk factors for recurrence of the VKH syndrome were investigated statistically by using a multiple logistic regression model. The clinical data of 87 patients were used for the statistical analysis. Among them, 58 patients (66.8%) had no recurrences, while 29 patients (33.3%) had recurrences. By multiple logistic regression analysis, dysacousia (p<0.01), cutaneous manifestations (p<0. 05), prodromal symptoms (p<0. 05), onset to treatment interval (days) (p<0. 05) and retinal detachment (p<0. 01) were significantly and independently associated with the recurrence. The Relapse Score was constructed by using a logistic model as follows: Relapse Score=+1. 459X (Dysacousia) + 1. 458x (Cutaneous manifestations) +0. 032x (Onset to treatment interval) -1. 637x (Prodromal symptoms) -1. 773x (Retinal detachment)+1. 247. It is supposed that the Relapse Score might be helpful for predicting the clinical course and modifying the dose or duration of systemic steroid therapy of the patients with the VKH syndrome.