Follow-up modalities for adult coeliac patients remain controversial. Non-invasive markers to identify coeliac patients on a gluten-free diet (GFD) with persistence of villous atrophy (VA) are still lacking. We aim to develop a score to stratify coeliac patients on a GFD according to their risk of having persistent VA and to tailor follow-up modalities accordingly. The clinical notes of over 700 coeliac patients attending our unit (September 1999–November 2018) were retrospectively examined. A total of 273 patients on a GFD with a histological follow-up performed 12–24 months after diagnosis were selected. We developed a bivariable model based on diet adherence and clinical response evaluated by previously validated methods. A four-level score (0·5, 1·5, 3, 4) was obtained. Patients on a strict GFD and with good clinical conditions (score 4) have a very low risk of persistence of VA (2 (95 % CI 1, 5) %). Conversely, the risk is very high (46 (95 % CI 25, 68) %) in patients with poor adherence to a GFD and unsatisfactory clinical response (score 0·5). A score of 1·5 (poor GFD adherence and persistent well-being) is linked with a high risk (23 (95 % CI 14, 36) %). Risk is intermediate (6 (95 % CI 3, 10) %) in patients scoring 3 (strict GFD and no/partial clinical improvement). Three patients who developed complications belonged to this scenario. Patients at low risk of persistent VA can be followed-up non-invasively, whereas a biopsy should be repeated in those at high/very high risk. Case-by-case evaluation is needed in patients at intermediate risk. Studies on a larger sample size are required to confirm these data.