In Saccharomyces cerevisiae, spreading of the telomeric SIR heterochromatin complexes into centromere-proximal euchromatic regions is prevented by the activity of boundary elements. So far, these boundaries have been associated with chromatin opening activities, like histone acetyltransferases (HATs) or histone methyltransferases. Here, we show that the opposite enzymatic activity, the histone deacetylase (HDAC) Rpd3, was necessary to prevent the encroachment of heterochromatin into euchromatin at telomeres in S. cerevisiae. We found by ChIP analysis that in the absence of Rpd3, the SIR complexes were mislocalized to more centromere-proximal regions, showing that Rpd3 was necessary to restrict SIR complexes to the telomere. Furthermore, quantitative RT-PCR showed that SIR proteins repressed subtelomeric genes in rpd3Δ cells, suggesting a role for Rpd3 in the restriction of telomeric heterochromatin. When combined with the absence of the known boundary factor, the HAT SAS-I, rpd3Δ caused inappropriate SIR spreading that was lethal to yeast cells. Significantly, the lethality between sas2Δ rpd3Δ was suppressed by sir deletions, suggesting parallel functions for the two enzymes in restricting SIR proteins to heterochromatin despite their opposing enzymatic activity. In addition, Rpd3 was capable of creating a boundary when targeted to the heterochromatic loci, indicating a boundary function for Rpd3. Further analysis showed that Rpd3 in essence functioned by removing acetyl groups, such that they were no longer available for NAD+-dependent deacetylation via Sir2. This further suggested that prevention of O-acetyl-ADP-ribose (OAADPR) production during deacetylation by Sir2 in effect prevented SIR propagation. This hypothesis was strengthened by the notion that inhibition of OAADPR binding to Sir3 created a halt to SIR spreading. In further experiments, we found that Rpd3 interacted in vivo with Cac1, the largest subunit of the chromatin assembly complex CAF-I, suggesting that it deacetylated cytoplasmic histone acetylation marks in a replication-coupled fashion. Thus, Rpd3 likely performed its function in SIR restriction through a transient contact to chromatin, rather than being permanently located at subtelomeric regions. In summary, our data indicated that Rpd3 effectively removed acetyl groups in subtelomeric regions and therefore deprived Sir2 of its ability to perform the deacetylation reaction and in doing so to produce OAADPR. This in essence prevented SIR propagation and created a boundary against heterochromatin spreading.