Numerous cancer fusion genes have been identified and studied, and in some cases therapy or diagnostic techniques have been designed that are specific to the fusion protein encoded by the fusion gene. However, there has been little progress in understanding the general features of cancer fusions genes in a way that could provide the foundation for an algorithm for predicting the occurrence of a fusion gene once the chromosomal translocation points have been identified by karyotype analyses. In this study, characterization of 59 cancer fusion genes indicated that all but a small percentage of the genes involved in fusion events are either relatively large, compared to neighboring genes, or are highly conserved in evolution. These results support a basis for designing algorithms that could have a high degree of predictive value in identifying fusion genes once conventional microscopic analyses have identified the chromosomal breakpoints.