Abstract Aims To assess the protective effect of statins in a large and unselected cohort of frail elderly subjects. Methods and results The 460 460 Lombardy residents (Italy), aged ≥65 years, who received ≥3 consecutive prescriptions of a statin during 2011–2012 were identified. A case–control study was performed, the cases being the cohort members who died during 2011–2018. Logistic regression was used to model the outcome risk associated with statin adherence. Adherence to drug therapy was measured by the proportion of the follow-up covered by prescriptions. The analysis was stratified according to four clinical categories (good, medium, poor, and very poor clinical status), based on different life expectancies, as assessed by a prognostic score which had been found to sensitively predict the risk of death. The 7-year death probability increased from 11% (good) to 52% (very poor clinical status). In each clinical status, there was a significant reduction of all-cause mortality as adherence to statin treatment increased. The reduction in the adjusted risk of mortality from the lowest to the highest adherence level was greatest among patients with a good clinical status (−56%) and progressively less among other cohort members, i.e. −48%, −44% and −47% in medium, poor, and very poor groups, respectively. Similar findings were obtained for the risk of cardiovascular mortality. Conclusion In a real-life setting, adherence to statin treatment reduced the death risk also in frail elderly patients. However, in these patients, the benefit of statin treatment may be lower than in those in good clinical conditions.