The message to take away from much of the research into obesity is "don't let it happen to you." Fortunately for most of us, that is well within our power: obesity is almost always a choice.
While smoking reduces life by an average of ten years, the research says being seriously overweight can cut life expectancy by as much as 13 years.
As for any complicated survey of human health, you'll find dissenting studies, or studies that show widely different results for different studied groups. Here's one that shows no real difference to life expectancy for obese versus non-obese old people:
Total, active, and disabled life expectancy in Americans aged >/=70 is estimated, with and without obesity and arthritis. Results indicate that neither obesity nor arthritis is related to the length of life for older men and women, alone or in combination. However, both conditions are significantly individually associated with increased length of disabled life in older men (1.4 years attributable to obesity; 1.2 years to arthritis at age 70) and women (1.7 years attributable to obesity; 2.1 years to arthritis at age 70). In addition, the combination of the two is significantly related to decreased active life, with nearly 50 and 60% of remaining life for 70-year-old men and women lived with disability, respectively.
I suspect there's some survivorship bias in studying people who are sufficiently inured to obesity by luck of genes or more active lifestyle to make it past 70 - despite the nasty effects that all that visceral fat has on your metabolism. Even if this you're not losing years, it certainly takes its toll in other ways. Disability and age-related disease of any sort is an unpleasant, very expensive experience, yet many people ensure they will suffering more of it by way of the lifestyle they lead and the calories they consume.