Some statistical research information
here.
Makes some interesting reading.
There are a couple of other very good threads currently running elsewhere on PPRuNe regarding single and twin engine safety that are worth reading.
Rotorheads view
Dunnunder view
As engine failure rates are calculated solely on the ratio of failures per 100,000 flight hours, the statistical probability of a
single engine failure on a twin are in fact doubled. In the wrong machine at the wrong time, your prospects are very limited.
Chimbu, you're experiences there make very interesting reading.