When a species is evolving by diverging or isolation, at one point they will end up practically unable to cross-breed. At that point, they will have become separate species.
Except what
Jadzia and I are positing is this is unlikely to be an overnight thing - one generation can breed to produce fertile offspring, the next simply can't - except perhaps in rare cases resulting from a mutation that directly affects breeding. Odds are, we're usually talking a decline of probabilities over a number of generations, and a distinct 'grey area' when it comes to defining speciation. After all, the definition of species is more than just the breeding issue, it is a definition which is controversial even in its own field as that deceptively simple approach ignores a great many subtleties which turn out to be quite important (this thread has thrown up a few already).