The X-linked recessive gene hypothesis, a theory to predict the mode of genetic inheritance of spatial ability, was tested using a new Rasch-calibrated space test, the Three-dimensional Cube Test (Gittler, 1990). This allowed solving the homogeneity problem (insufficient unidimensionality of test material) by which earlier research in this field was affected. The empirical correlations from our sample of 134 families were compared with the theoretically predicted values. Present data do no corroborate the model; however, the problems of its general falsification are discussed. The fact that differences in performance in favour of males exist in the parental generation but not in the filial generation accentuates the importance of environmental factors.