**Introduction :** IMAGEN consortium data from adolescents at age 14 were used for binary prediction analyses of future clinical anxiety vs. future healthy control between the ages of 18 and 23, using psychometric features only, neurostructural features only, or both combined. For each featureset, one analysis has been conducted using all participants with future anxiety (All, N = 156), one only with participants with future GAD and no other anxiety diagnosis (GAD, N = 42), and one with participants with future multiple anxiety diagnoses (mAD [at least two distinct anxiety diagnoses], N = 42). All participants with future anxiety did not have an anxiety diagnosis at the time baseline data was collected. All prediction analyses use a N = 424 healthy control group that did not have any diagnosed psychiatric disorder at ages 14, 18 nor 23. **Detailed content :** Each folder contains pickled trained classifiers of all 56 cross-validation iterations, as well as the name and order of the features used, and the global ROC curve.