Fabry disease is a genetic disorder caused by a deficiency of alpha-galactosidase, exhibiting a wide clinical spectrum, from the early-onset severe 'classic' form to the late-onset mild 'variant' one. Recent screening of newborns revealed that the incidence of Fabry disease is unexpectedly high, and that the genotypes of patients with this disease are quite heterogeneous and many novel mutations have been identified in them. This suggests that a lot of Fabry patients will be found in an early clinical stage when the prognosis is obscure and a proper therapeutic schedule for them cannot be determined. Thus, it is significant to predict the clinical phenotype of this disease resulting from a novel mutation. Herein, we proposed a phenotype prediction model based on sequential and structural information. As far as we know, this is the first report of phenotype prediction for Fabry disease. First, we investigated the sequential and structural changes in the alpha-galactosidase molecule responsible for Fabry disease. The results showed that there are quite large differences in several properties between the classic and variant groups. We then developed a phenotype prediction model involving the decision tree technique. The accuracy of this prediction model is high (86%), and Matthew's correlation coefficient is also high (0.49). The phenotype predictor proposed in this paper may be useful for determining a proper therapeutic schedule for this disease.