Does it depend on the phase in which it has been blocked, i.e. G0/G1 will take one time while G2/M will have another? Does it depend on the cell's type (dividing cells) or the cells lines one is looking at?
Firstly some data from a setting without chemical synchronisation: I observed individual HeLa cells over prolonged times. On average, sibling cells become asynchronous in their next division by about 80 min (>160 cell pairs analysed) - Rehm M et al, Cell Death Differ 2009.
For phase-based synchronisation you'll need to read this paper: Cooper S. Rethinking synchronization of mammalian cells for cell cycle analysis. Cell Mol
Life Sci 2003; 60: 1099–1106.
This really well discusses the implications of different phase-arrests with respect to the synchrony that can be achieved.
Generally, one cell cycle is a reasonable assumption. However, as Jonathan mentioned this can be highly cell specific. If the cell line has a mutation in cell cycle checkpoint proteins or so they will leak. Here attached is a provocative view of cell synchronization.