Cancer is a complex family of diseases. From the viewpoint of molecular biology, cancer is a genetic disease resulting from abnormal gene expression due to DNA instability, such as translocation, amplification, deletion or point mutations. In this paper, the proposed approach is based on functional domain analysis of the cancerrelated genes to generate a set of rules for prediction of another set of genes that are also cancer-related. An assessment indicated that a total accuracy of 81.45% was reached. The satisfactory result demonstrates the robustness and usefulness of the system.