Two paradoxes of centromere biology have confounded our understanding of the eukaryotic centromere, and consequently of chromosome evolution. The frist is that centromere function is highly conserved across eukaryotes, yet centromere-specific proteins that interact with nucleic acids diverge rapidly. The second is that while satellite DNA is found ubiquitously across eukaryotic centromeres, it is considered neither necessary nor sufficient for centromere formation. Major hurdles in understanding centromere evolution lie in the highly repetitive nature of most centromeric DNA and an inability to decouple centromere divergence from species evolution and stochastic processes such as genetic drift and molecular drive. Using comparative cytogenomics, we have identified specific small noncoding RNA sequences that are coincident with active centromere demarcation in a broad range of mammalian species spanning all Therian clades, affording the ability to assess the evolution of centromeres in the context of the transcriptional activity of nascent centromeric elements. As an exemplar, the macropodid species complex is typified by rapid chromosome evolution and convergence of karyotypes independent of ancestry. Notably, each of these karyotypic rearrangements involves centromeres and hybrids between even closely related species possess abnormalities delimited to centromeres and follow the observations of Haldane’s Rule, indicative of strong, postzygotic reproductive isolation barriers. Our ChIP-seq, RIP-seq, RNA-seq, genome assembly and repeat analysis computational methods have established a testable model of centromere evolution in the context of rapid chromosome and species evolution. Our findings on satellites, retroelements and noncoding RNA elements will be presented in the context of conflict between nucleic acid binding proteins and centromeric DNA domains, and utlimately chromosome evolution.