I occaisionally use the monte carlo analysis in Microwave Office. It has the facility to tell you what the input states were for each output, so if you've got a minor spread problem you can see what's causing the issue.
Mostly Monte Carlo is used to get a design past a review stage where the reviewer isn't that competent but likes seeing pretty pictures.
As far as on chip resistors go, those of a similar value will be fairly closely matched, so an R2R ladder on a DAC can be reasonable without having to laser trim.