This paper aimed to test the specificity of predicting power of finger gnosia on later numerical abilities in school-age children and to contribute to the understanding of this effect. Forty-one children were tested in the beginning of Grade 1 on finger gnosia, left-right orientation (another sign of the Gerstmann "syndrome"), and global development. Fifteen months later, numerical and reading abilities were assessed. Analyses of the results indicated that, contrary to the general measures of cognitive development, performance in the finger gnosia test was a good predictor of numerical skills 1 year later but not of reading skills, which proves the specificity of that predictor. The same conclusion was also true for the left-right orientation. However, finger gnosia could equally predict performance in numerical tasks that do or do not rely heavily on finger representation or on magnitude representation. Results are discussed in terms of the localizationist and the functional hypotheses.