MeCP2 is an abundant methyl-cytosine-guanine (CG)-binding protein and transcriptional repressor. We developed a biochemical system that exhibits CG methylation-specific transcriptional repression by purified human MeCP2. MeCP2 represses transcription by histone deacetylase (HDAC)-dependent and HDAC-independent mechanisms. Our system appears to recreate the HDAC-independent component of MeCP2-mediated repression and occurs via inhibition of the assembly of transcription preinitiation complexes. At a ratio of approximately one molecule of MeCP2 per two methyl-CG dinucleotides, as found in mammalian neurons, the magnitude of methylation-specific repression was greater than 10-fold. Notably, the HDAC inhibitor trichostatin A had no effect on MeCP2-mediated repression with either naked DNA or chromatin templates. We designed a CG-deficient core promoter that is resistant to MeCP2-mediated repression when placed in a plasmid lacking CG dinucleotides. By using this CG-deficient reporter as a reference, we found that eight CG dinucleotides in the core promoter region are sufficient for strong methylation-specific repression by MeCP2. In contrast, MeCP2 does not repress a construct with 13 CG dinucleotides located ∼1.7 kbp upstream of the promoter. Furthermore, by analysis of C-terminally truncated MeCP2 proteins, we found that binding of MeCP2 to methyl-CG dinucleotides is not sufficient for transcriptional repression. Hence, MeCP2-mediated repression is not due to the simple steric blockage of the transcriptional machinery. These experiments suggest that MeCP2 can function as a global methyl-CG-specific, HDAC-independent repressor. This HDAC-independent mechanism of MeCP2-mediated repression may be important in cells, such as mammalian neurons, that have high levels of CG methylation and MeCP2.