The mechanisms that specify precisely where mammalian kinetochores form within arrays of centromeric heterochromatin remain largely unknown. Localization of CENP-A exclusively beneath kinetochore plates suggests that this distinctive histone might direct kinetochore formation by altering the structure of heterochromatin within a sub-region of the centromere. To test this hypothesis, we experimentally mistargeted CENP-A to non-centromeric regions of chromatin and determined whether other centromere-kinetochore components were recruited. CENP-A-containing non-centromeric chromatin assembles a subset of centromere-kinetochore components, including CENP-C, hSMC1, and HZwint-1 by a mechanism that requires the unique CENP-A N-terminal tail. The sequence-specific DNA-binding protein CENP-B and the microtubule-associated proteins CENP-E and HZW10 were not recruited, and neocentromeric activity was not detected. Experimental mistargeting of CENP-A to inactive centromeres or to acentric double-minute chromosomes was also not sufficient to assemble complete kinetochore activity. The recruitment of centromere-kinetochore proteins to chromatin appears to be a unique function of CENP-A, as the mistargeting of other components was not sufficient for assembly of the same complex. Our results indicate at least two distinct steps in kinetochore assembly: (1) precise targeting of CENP-A, which is sufficient to assemble components of a centromere-prekinetochore scaffold; and (2) targeting of kinetochore microtubule-associated proteins by an additional mechanism present only at active centromeres.