I am utterly baffled by the notion that a target number (option b) is more sensible. It may well be what you are used to, but in purely mathematical terms the percentile approach is more straightforward, and is the approach used in every simulation I've ever seen. This is because:
1. Most scientific data can be used to construct a (0,1) probabilty of something happening.
2. A large number of computerised systems are capable of generating good pseudo-random (0,1) numbers.
3. You may want to use a probability distribution to sample from, like the poisson or normal distribution.
In a completely made up doolalli gameworld like D&D there's no reason why points 1 and 3 would interest you. But in any remotely real world setting they might.
Kaft, if you're talking about game design then you may want to include these points, because it can greatly simplify making your target tables and so on.
Hence... *and with this Walsingham recedes into a tangle of anecdotes concerning risk analysis and, for some reason, baboons.*