Background: Adverse drug reactions (ADRs) in older people are often preventable, indicating that screening and prevention programs aimed at reducing their rate are needed in this population.Objective: The aim of this study was to externally validate the GerontoNet ADR risk score and to assess its validity in specific subpopulations of older inpatients. Methods: Data from the prospective CRIteria to assess appropriate Medication use among Elderly complex patients (CRIME) cohort were used. Dose-dependent and predictable ADRs were classified as type A, probable or definite ADRs were defined according to the Naranjo algorithm, and diagnostic accuracy was tested using receiver operating characteristic (ROC) analyses. Sensitivity and specificity were calculated for a cut-off point of 4. Results: The mean age of the 1075 patients was 81.4 years (standard deviation 7.4) and the median number of drugs was 10 (range 7–13). At least one ADR was observed in 70 patients (6.5%); ADRs were classified as type A in 50 patients (4.7%) and defined as probable or definite in 41 patients (3.8%). Fair diagnostic accuracy to predict both type A and probable or definite ADRs was found in subpopulations aged <70 or ≥80 years with heart failure, diabetes, or a previous ADR. Good accuracy to predict type A ADRs was found in patients with a low body mass index (BMI; >18.5 kg/m2) and a Mini-Mental State Examination (MMSE) score of >24/30 points, as well as in patients with osteoarthritis. The cut-off point of 4 points yielded very good sensitivity but poor specificity results in these subpopulations. Conclusion: This study suggests that the GerontoNet ADR risk score might represent a pragmatic approach to identifying specific subpopulations of older inpatients at increased risk of an ADR with a fair to good diagnostic accuracy.@en