They are measured in different ways. AWStats uses analyzed log logs, and they include crawlers and bots, as well as end users with JavaScript disabled and Google Analytics users, none of which matter in Google Analytics.
AWStats creates visits from a combination of views in server logs from their IP address, so they do not follow a user who visits multiple locations or has a dynamic IP address, and counts multiple users from the same IP address as one visitor. Google Analytics uses browser cookies to track visitors several times in several places. Both may have a tendency to inflate or deflate numbers. Thus, server logs can be counted by several people on the same network as the same person, but they also double the count when moving and have no idea how to handle dynamic IP addresses. Google Analytics cannot track a single user across browsers.
So the correct answer, in general, is that no analytics tracking will ever be 100%, that numbers should always be considered approximations, and that every number you look at should be viewed in the context of how its tracked and only compared to numbers collected in similar contexts. The general trend is that AWStats is exaggerating numbers and that Google Analytics is downplaying them, but this is not an obscene rule.
Yahel source share