I would assume that the Born-Oppenheimer approximation breaks down when the motion of the nuclei or the forces exerted upon them become nonnegligible. This could happen at very high temperatures (lots of kinetic energy) or if the nuclei are exceedingly light.
.
