Here's one plausible story: The '90s saw the advent of drugs like Pravachol and Lipitor that can dramatically cut your cholesterol and increase your life expectancy. With medical advances like that, who needs to be thin? Of course obesity is still bad for you—but it's not as bad for you as it used to be. The price of obesity (measured in health risks) is down, so rational consumers will choose more of it.
With the success of the human genome project, even greater advances are just over the horizon, making obesity an even greater bargain. Today's expanding waistlines might reflect nothing more than a rational expectation of future progress against heart disease.
If you don't like that story, here's another: The '90s were the era of low-fat foods. At fewer calories per serving, it makes sense to eat more servings. The net effect can be either an increase or a decrease in weight. For example: Suppose a scoop of ice cream a night would add 10 pounds to your weight, and you've decided that's not worth it, so you don't eat ice cream. Now along comes a low-fat ice cream that allows you to eat two scoops a night and add 10 pounds to your weight. That's a better deal, and a perfectly rational being might well opt for it. So when low-fat foods come along, some people sensibly decide to become fatter. (Other people, equally sensibly, use low-fat foods to become thinner. Therefore, the overall impact of low-fat foods on obesity could in principle go in either direction.)