Prior work on list-method directed forgetting (LMDF) has shown that instructions to forget previously encoded material are highly effective and can support memory updating. Forgetting instructions are however usually presented immediately after encoding of to-be-forgotten information. It is currently unclear if voluntary forgetting can also support updating when information was encoded some time ago and, arguably, already stabilized in memory. Two experiments applied the typical LMDF task to address this gap in the literature. Participants studied two lists of unrelated items and, between lists, received instructions to either remember or forget the first list. Critically, the retention interval after list-1 and before list-2 encoding was varied, with participants receiving remember or forgetting instructions after shorter retention intervals (0-20 min) or after a prolonged retention interval (24 hrs). The results showed intact voluntary forgetting after the shorter retention intervals, but not after a prolonged retention interval of 24 hrs. This suggests that it may be harder to forget information that was already stabilized in memory. Prolonged retention intervals could thus act as a boundary condition for effective memory updating in the form of LMDF.