ObjectivesTo assess the use of MRI-determined tumour regression grading (TRG) in local response assessment and detection of salvageable early local relapse after chemoradiotherapy (CRT) in patients with anal squamous cell carcinoma (ASCC).MethodsFrom a prospective database of patients with ASCC managed through a centralised multidisciplinary team, 74 patients who completed routine post-CRT 3- and 6-month MRIs (2009â2012) were reviewed. Two radiologists blinded to the outcomes consensus read and retrospectively assigned TRG scores [1 (complete response) to 5 (no response)] and related these to early local relapse (within 12 months) and disease-free survival (DFS).ResultsSeven patients had early local relapse. TRG 1/2 scores at 3 and 6 months had a 100 % negative predictive value; TRG 4/5 scores at 6 months had a 100 % positive predictive value. All seven patients underwent salvage R0 resections. We identified a novel âtram-trackâ sign on MRI in over half of patients, with an NPV for early local relapse of 83 % at 6 months. No imaging characteristic or TRG score independently prognosticated for late relapse or 3-year DFS.ConclusionsPost-CRT 3- and 6-month MRI-determined TRG scores predicted salvageable R0 early local relapses in patients with ASCC, challenging current clinical guidelines.Key Points⢠Post-chemoradiotherapy MRI (3 and 6 months) helps local response assessment in ASCC.⢠The MRI-TRG system can be used reproducibly in patients with ASCC.⢠The TRG system facilitates patient selection for examination under anaesthesia and biopsy.⢠The use of MRI-TRG predicts for detection of salvageable early local relapses.⢠The TRG system allows for a standardised follow-up pathway.Electronic supplementary materialThe online version of this article (doi:10.1007/s00330-016-4337-z) contains supplementary material, which is available to authorized users.