BACKGROUND: Recent evidence suggests that both preoperative and postoperative inflammation-based prognostic markers are useful for predicting the survival of colorectal cancer (CRC) patients. However, associations between longitudinal changes in inflammation-based prognostic markers and prognosis are controversial. METHODS: The subjects of this study were 568 patients with stage III CRC between 2008 and 2014. Preoperative and postoperative neutrophil-to-lymphocyte ratio (NLR), lymphocyte-to-monocyte ratio (LMR), C-reactive protein/albumin ratio (CAR) and lymphocyte-to-C-reactive protein ratio (LCR) were calculated to assess the inflammatory state of subjects. Subjects were stratified into three groups for each marker: preoperatively low inflammatory state (normal group), preoperatively high but postoperatively low inflammatory state (normalised group) and persistently high inflammatory state (elevated group). Multivariable analyses for overall survival (OS) and recurrence-free survival (RFS) were performed to adjust for well-established clinicopathologic factors. RESULTS: For all assessed markers, the normalised group had a significantly better prognosis than the elevated group and a similar prognosis as the normal group for both OS and RFS. CONCLUSIONS: Postoperative, but not preoperative, inflammation-based prognostic markers more accurately predict OS and RFS in patients with stage III CRC.