Genetic evidence suggests that the securin Pds1p is the target of a late-S-phase checkpoint control. Here we show that Pds1p becomes essential once two-thirds of the genome has been replicated and that the coupling of the completion of genome replication with mitosis relies on the regulation of Pds1p levels. Mec1p is needed to maintain Pds1p levels under S-phase checkpoint conditions. In contrast, Rad53p and Chk1p, needed for the stabilization of Pds1p in the context of the G2 DNA-damage checkpoint pathway, are dispensable. Thus, the Pds1p-dependent late-S-phase checkpoint pathway couples replication with mitosis but is mechanistically distinct from the G2 DNA-damage checkpoint. Finally, we show that the inhibition of spindle elongation in early S phase, controlled by the Mec1p/Rad53p branch, is not regulated via Pds1p/Esp1p. This can mechanistically explain the need for branched S-phase checkpoint controls.