Measuring instruments for behavioral addictions tend to not be generalizable to other dependences. The aim is to carry out a process of adaptation of the General Addiction Scale (EAG) (Ramos, Sansebastian & Madoz, 2001), to assess exercise dependence. This test presents a self-administered one-dimensional character capable of measuring the degree of addiction of a subject to all kinds of new addictions, excluding illicit substances of abuse. The scale, after adjustment, was renamed as Training Addiction Scale (EAE). 401 athletes took part of which 82.29% are men and the whole sample indicated an age with M=45.78 and SD = 10.25 years. The confirmatory factor analysis has allowed discriminating a general factor and four subscales, with Cronbach's alpha for each of the sub-scales: Tolerance (α=.78), pleasure-relaxation (α=.77), Lack of Control (α=.77) and Abstinence-craving (α=.71). The indexes of asymmetry and curtosis have been near zero and <2.0. The items showed no overlap between the subscales. The model presented correct values for determining an acceptable goodness of fit of the original model and the results were: χ2 / df = 2.93, IFI = .96, CFI = .96, TLI = .93, SRMR = .039, RMSEA = .049.