I would apply some economic metrics here:
In order to consider an evidence, I would estimate the cost of forging this evidence first.
For example, before paying their mercenaries, paramilitary organizations demand video-recordings of the hit-and-run operations. Of course, one could hire Hollywood to produce such a recording, but they -- correctly -- figure that it's cheaper for the thugs to actually do the dirty work instead of producing a fake recording.
Similarly with emails: there are costs associated with forging a digital signature and there are costs associated with faking an unsigned message (depending on what information is in there). It depends on the actual case, how these costs measure up to the claims.
I would think that in the digital world, where anything can be forged and nothing can be trusted with absolute certainity, this cost-benefit approach to evidence is the only way to go for justice providers. Be it criminal or civil cases, state-certified courts, or out-of-court arbitration services.