I have tested it both ways. In some cases, I only gain an extra 10% of urls found from using the internal link extractor. over what the engines give, while making me sort thru 200% more total urls to get that 10% gain.
While other times I have gained almost 100% increase over what the engines give. So it all depends on the site. This is working with about 5000 unique domains total that were all auto approve.
I always used the engines as a base. However with yahoo pulling the API and the site: not supported in scrapebox for yahoo, using the internal link extractor is way more appealing.
Also AOL yeilds a lot of site: results, but they are broken right now too, so right now I am using the internal link extractor with good results.
If you really wanted to do it, the ideal way would be to take your AA urls, then run them thru site: on all working engines, then run those results thru the internal link extractor.
I do like to include the engines though, that way you get only already indexed pages, where using just the internal link extractor may get you pages that are mostly not indexed. It just all varies, so I use both.