Objective: A subset of Crohn’s disease (CD) patients experiences long-term remission after infliximab withdrawal. Biomarkers are needed to identify those patients. Design: New biomarkers of relapse were searched in the baseline serum of CD patients stopping infliximab when they were under combined therapy (antimetabolite and infliximab) and stable clinical remission (STORI cohort, n=102). From shotgun proteomics experiment (discovery step), biomarker candidates were identified and further targeted by selected reaction monitoring (verification step). The dataset was stratified to search for markers of short- (<6 months) or long-term relapse (>6 months). The risk of relapse and the predicting capacity associated with biomarker candidates were evaluated using univariate Cox model and log-rank statistic, respectively. To test their complementary predicting capacity, biomarker candidates were systematically combined in pairs. Results: Distinct biomarker candidates were associated with the risk (hazard ratio: HR) of short- (15 proteins, 2.9<HR<16.1, p<0.05) and long-term (17 proteins, 2<HR<4.4, p<0.05) relapse, they reflect different pathophysiological processes. In stratified and non-stratified datasets, novel marker combinations exhibited a high predicting capacity as shown by their higher Z-scores (FDR<0.001) than CRP and faecal calprotectin (current references in predicting relapse). Conclusion: We identified for the first time circulating biomarker candidates associated with the risk of long-term relapse in CD patients stopping infliximab. We also highlight a sequence of pathophysiological processes leading to relapse, this could help to better understand the disease progression. Our findings may pave the way for a better non-invasive evaluation of the risk of relapse when contemplating anti-TNFα withdrawal in CD patients.