No reliable identification of quiescent Crohn's disease (CD) patients with a high risk of relapse is available. The aim of this study was to develop a prognostic index to identify those patients. Untreated adult patients with quiescent disease (not induced by surgery) included in three phase III clinical trials were analysed retrospectively with respect to time to relapse. Nineteen factors related to biology, disease history, and topography were investigated. A relapse was defined as either a CD Activity Index (CDAI) > or = 200, a CDAI > or = 150 but over the baseline value by more than 100, or acute complications requiring surgery. The inclusion criteria were fulfilled by 178 patients. The median follow up was 23 months. The Cox model retained the following bad prognostic factors: age < or = 25 years, interval since first symptoms > 5 years, interval since previous relapse < or = 6 months, and colonic involvement (p < 0.001). Bootstrapping confirmed the variable selection. Patients were classified into three groups with an increasing risk of relapse (p < 0.001). The worst risk group was composed of patients presenting at least three of the four bad prognostic factors. These results make possible the design of clinical trials in quiescent CD patients with a high risk of relapse.