ROCHESTER — If you're one of those people who like to track all the calories you just burned on the treadmill, you might want to sit down for this. You're going to need to start cutting that number in half.

Unless you're running really hard. In that case, you'll need to cut it by two-thirds.

That's one of the findings of E-MECHANIC, a new, five-year study from one of the premier obesity research labs in the US. The name of the paper, recently published in the American Journal of Clinical Nutrition, is short for "Examination of Mechanism of Exercise-Induced Weight Compensation," an NIH-funded attempt to learn why exercise alone does not seem bring about weight loss.

"When people exercise, they tend to lose less weight than you would expect, based on the increased energy expenditure from the exercise," says Corby Martin, a professor at Pennington Biomedical Research Center in Louisiana and lead author of the study. "If you compute how much weight you should lose based on how many calories you burn from the exercise, people usually lose less than half of that. And of course this is really quite frustrating. It made us question why that was occurring."

Researchers have long known the calories burned in an exercise program do not translate in a one-to-one relationship to pounds lost. It is one of those unwelcome caveats to training that has made itself visible over time. What they had never attempted to figure out was the exact reason why.

Did people new to training hit the couch a little harder once they got home ? Did their metabolism slow to a crawl? Or did the answer to this mystery of calories not burned lie in that pile of energy bar wrappers over there in the gym bag?

Beginning in 2010, Martin and his colleagues at Pennington recruited nearly 200 subjects. They randomly assigned 65 of them to a do-nothing control group, 66 to a half-year of moderate aerobic exercise, and 67 more to a rigorous six months of treadmill runs. They used a host of calculations to track of how much energy these subjects were burning on their time off, how many calories they were eating, how many they were working off in the gym and whether they had slowed down during the rest of their day.The participants were given no dietary instructions.

As expected, no one lost as many calories as they burned.

"The group that did a little exercise," Martin says, "only got about 40 to 45 percent of the benefit. The group that did a lot of exercise only got about a third of the benefit...For many individuals in this project, they were really frustrated, because they swore their food intake wasn't changing. They were exercising a lot, they weren't losing what they expected, and they were frustrated because they didn't understand how this was happening."

The results showed the exercisers' metabolism was fine, thank you very much, and they were not suddenly getting stationary around the house. They were eating more, however, especially if they were working out the hardest.

"Compared to controls," Martin says, "the two exercise groups had an increase in food that averaged about 90 to 125 calories a day." That's not a lot of food, he points out.

"On any given day we might eat an extra 200-300 calories, but the next day we might eat 200-300 less, so it averages out over time and in the end we are pretty weight stable. In this case, it was a pretty consistent increase of 100 to 125 calories, so over the course of that six months, that 100-125 calories per day really did diminish the amount of weight they lost."

Martin says the extra calories were coming during snacks and sweetened drinks, as opposed to lunch and dinner, and were so small as to fall below the level of awareness. "This increase really appears to be very insidious. An extra 100 to 125 calories a day, that's not much. It's really hard to detect. It's an extra bite or two of peanut butter, or an extra apple and a half."

The exercisers who "compensated" in this fashion also tended to endorse a common refrain of exercisers everywhere: I worked out, therefore I deserve a little something extra. (They may also have been hungrier.) That said, "they didn't believe their food intake had changed."

"Weight loss is really challenging, period," says Michael Joyner, professor of physiology at Mayo Clinic. "The only method that's been shown to work, with diet and exercise, is adherence. "

Whatever your dietary and exercise prescription, Joyner adds, it has to be sustainable. " I try to build low-level activity into my day, to get 40 minutes to an hour of structured exercise most days. I try to be mindful what I eat. I try to avoid certain foods. Many people learn to eat the same thing for lunch and breakfast every day, and just reduce all the choices."

Martin says nothing about E-MECHANIC should be taken as advice to drop exercise, which has a host of benefits unrelated to weight loss. "But these data show that when you exercise, while you of course you lose some weight, it's not quite as much as it should be. So if weight loss is your goal, then you should also focus on diet."