And of course it all depends on what voltage is considered significant, ie 'not ground.'
http://science.nasa.gov/science-news/science-at-nasa/lis/lis_4/"The voltage difference between ground and ionosphere is 200,000 to 500,000 volts (200 to 500 kV)."
Note that the ionosphere is a conductive medium, so is the 'ground', and the atmosphere between isn't a very reliable insulator, especially when wet. Hence lightning.
Also there are flavors of 'ground.' Chassis ground, circuit ground (which may or may not be connected in some way to chassis ground), shield ground (ditto), summing ground points, etc. Ground isn't necessarily fixed either, hence 'ground bounce' as referred to in logic design. (It's undesirable btw.) While in RF there's no such thing as 'ground' since everything is a circuit element with complex impedance - even the 'ground' plane. While in waveguide systems, where is your ground now? Or there are 'grounds' that are so localized in a circuit that people use a different word - 'common', eg "common emitter transistor circuit" where the reference is just one circuit node, but it's still a ground of sorts.
Then there's high voltage discharge devices, like can crushers, traveling arc lasers, and so on. In which the entire circuit is 'ground' more or less, but with extreme transient effects.
'Ground' is just a word, that can mean many different things depending on the context. Ideally it's a circuit-wide reference, but wherever there's current flowing and non-perfect conductors, there's a voltage difference.
I sometimes think that understanding 'ground' is about 50% of understanding all of electronics.