Definition of a 'tunneling lifetime'

I'm given a one-dimensional potential with two wells, one local minimum at some higher energy and one deep global minimum next to it, separated by a barrier of own shape and height (phase qubit). I used a WKB approx. in conjunction with a harmonic approximation of the barrier to find the tunneling probability, i.e. the transmission coefficient T.

Now, I'm asked to estimate the tunneling lifetime of the metastable levels in the elevated well. How do you relate the tunneling probability to the lifetime?

-
Think of things which decay... –  Michael Brown Jul 10 '13 at 8:27
I looked into alpha decay where they set the decay constant equal to $vT/R$ with velocity $v$ and nuclear radius $R$. This seems to be very heuristically motivated with the image of having the alpha moving through the well and bouncing off the the barrier with prob. 1-T and tunneling through with prob. T. I wonder if that's the sort of semi-classicality that comes with WKB or whether there exists some rigorous quantum derivation. –  Jonas Jul 10 '13 at 9:13
There is a discussion in this reference Chapitre 2.3 Page 39 –  Trimok Jul 11 '13 at 10:01
Thanks, very helpful. –  Jonas Jul 11 '13 at 11:49