BACKGROUND: Antibodies to infliximab [ATI] and trough levels to infliximab [TRI] are associated with loss of response in inflammatory bowel diseases [IBD]. The best way to predict loss of response [LOR] to infliximab [IFX] is unknown. METHODS: We conducted a prospective observational cohort study enrolling all IBD patients who were in clinical remission at Week 14 after IFX treatment initiation. TRI, ATI and C-reactive protein [CRP] level were measured at Week 22 [T1] and thereafter at every other IFX infusion. Loss of clinical response was defined by a flare requiring therapeutic change [IFX dose intensification, initiation of another drug class, and/or surgery]. RESULTS: A total of 93 patients [59 Crohn's disease, mean duration of follow-up 17.2 months] were included; 32 patients [34.4%] lost clinical response during follow-up. Cumulative probability of LOR was 50% at 20 months. Mean TRI at T1 was significantly lower in IBD patients with stable ATI as compared with those with transient ATI or without ATI [0.052, 3.34 ,and 4.29 µg/ml, respectively; p = 0.001 between no ATI vs stable ATI, and p = 0.005 between stable and transient ATI] [p = 0.0001]. Three independent factors were predictive of LOR after Cox proportional hazards modelling: TRI > 5.5 µg/ml (hazard ratio [HR]: 0.21; 95% confidence interval [CI]: 0.05-0.89;p = 0.034) at T1, CRP > 5mg/l [HR: 2.5; 95% CI: 1.16-5.26; p = 0.019] at T1, and stable ATI defined by two consecutive ATI > 20ng/ml [HR: 3.77; 95% CI: 1.45-10.0; p = 0.007]. Transient ATI did not influence LOR. CONCLUSIONS: LOR can be predicted based on a combination of CRP, TRI and stable ATI with a high degree of accuracy.