Probiotics have been proposed as a therapy for inflammatory bowel disease, but variations in strains, formulations, and protocols used in clinical trials have hindered the creation of guidelines for their use. Thus, preclinical insight into the mechanisms of specific probiotic strains and mode of administration would be useful to guide future clinical trial design. In this study, live, heat inactivated (HI), and spent culture medium preparations of the probiotic Bifidobacterium breve NCC2950 were administered to specific pathogen free C57BL/6 mice before or during colitis, as well as before colitis reactivation. Five days of 3.5% dextran sulphate sodium in drinking water was used to induce colitis. Pretreatment with live B. breve reduced disease severity, myeloperoxidase activity, microscopic damage, cytokine production, interleukin (IL)-12/IL-10 ratio, and lymphocyte infiltration in the colon. B. breve did not attenuate on-going colitis. After acute colitis, disease symptoms were normalised sooner with live and HI B. breve treatment; however, reactivation of colitis was not prevented. These findings indicate that the efficacy of a probiotic to modulate intestinal inflammation is dependent on the formulation as well as state of inflammation when administered. Overall, live B. breve was most efficacious in preventing acute colitis. Live and HI B. breve also promoted recovery from diarrhoea and colon bleeding after a bout of acute colitis.