In vocational retraining centres, a test battery including several performance tests is generally administered to assess the occupational aptitude of rehabilitants and to predict their success in occupational retraining. This paper presents the multiple regressions of a set of achievement scores on "grades of retraining" and "grades of final examination", respectively, concerning retraining to become an office management assistant. It was shown that only few variables are adequate to clarify a maximum of variance of the criterions. Four different regression models were identified; each of them could clarify about 25% of variance. Significant predictors were indicators for verbal skills and basic numeracy. In each model a measurement for nonverbal intelligence had to be taken into account as a suppressor variable. To Put all in a nutshell, in order to predict the success of vocational retraining to become an office management assistant it is more important to focus on strengthening school knowledge than on general intelligence.