The replication clamp PCNA is loaded around DNA by replication factor C (RFC) and functions in DNA replication and repair. Regulated unloading of PCNA during the progression and termination of DNA replication may require additional factors. Here we show that a Saccharomyces cerevisiae complex required for the establishment of sister chromatid cohesion functions as an efficient unloader of PCNA. Unloading requires ATP hydrolysis. This seven-subunit Ctf18-RFC complex consists of the four small subunits of RFC, together with Ctf18, Dcc1, and Ctf8. Ctf18-RFC was also a weak loader of PCNA onto naked template-primer DNA. However, when the single-stranded DNA template was coated by the yeast single-stranded DNA binding protein replication protein A (RPA) but not by a mutant form of RPA or a heterologous single-stranded DNA binding protein, both binding of Ctf18-RFC to substrate DNA and loading of PCNA were strongly inhibited, and unloading predominated. Neither yeast RFC itself nor two other related clamp loaders, containing either Rad24 or Elg1, catalyzed significant unloading of PCNA. The Dcc1 and Ctf8 subunits of Ctf18-RFC, while required for establishing sister chromatid cohesion in vivo, did not function specifically in PCNA unloading in vitro, thereby separating the functionality of the Ctf18-RFC complex into two distinct paths.The process of sister chromatid cohesion ensures that replicated chromosomes are distributed equally to progeny cells during cell division. Cohesion is mediated through a large ring-like structure, cohesin, which is deposited on the chromosomes in the late G 1 phase of the cell cycle. Sister chromatid cohesion is established during S phase, presumably by the actual passage of the replication fork. Several models exist by which cohesin is proposed to interact with the duplicated chromosomes in order to mediate cohesion. These models are based either on the idea that one cohesin ring encircles both sister chromosomes or on the idea that cohesion is mediated by the interlocking of two cohesin rings, with each ring more peripherally associated with a daughter chromosome (reviewed in references 33, 34, and 48). If the first type of model is correct, the mere passage of the replication fork through a cohesin ring would invariably ensure that the two sister chromatids stay attached until mitosis. However, the disadvantage of this elegant solution to the chromosome sorting problem is that it may be problematic for the replication fork to pass through the estimated 30-to 40-nm hole of the cohesin ring. Recent studies have shown that cohesin is specifically redistributed along the chromosome to transcription termination sites in a transcription-dependent manner, suggesting that the actual transcription machinery may push the cohesin rings ahead of the transcription bubble (26). If this is caused by steric problems because of the size of the transcription apparatus, similar steric problems may also occur with passage of the replication fork. Failure to establish sister chromatid cohesion would resul...