Coffee and type 2 diabetes
This meta-analysis finds that increased coffee consumption is associated with a lower risk of type 2 diabetes. Remember that meta-analysis is just going back and combining multiple smaller studies into one big one to improve statistical power and that type 2 diabetes is the kind that tends to affect the old, the obese and the inactive. Risk was reduced about 1/3 for the highest coffee consumers (>6 or 7 cups per day, which seems like a lot), and 1/4 for those drinking a bit less.
I have heard it said that "meta-analysis is to analysis as metaphysics is to physics" which probably overstates the case, but meta-analysis does have some problems. In addition, these kind of studies can only show correlation, not causation.
Most of the studies aggregated into the meta-analysis corrected for factors like age, sex and obesity (you could imagine that the obese drink less coffee because they are too lazy to get up and get it), so the authors hypothesize a direct effect of some component of coffee (apparently not caffeine itself since decaf gives a similar benefit) on the insulin/glucose metabolism machinery.
I also learned from reading this that drinking filtered coffee decreases LDL (the bad cholesterol) compared to drinking pot-boiled coffee. Apparently the switch from pot-boiled to filtered in Finland is thought to have signficantly decreased the incidence of coronary artery disease.