Our bodies seem to adjust to prolonged, repeated physical exertion and its energy demands by burning fewer — instead of more — calories over the course of the day, even if our exertions continue at the same level, according to a surprising new study of energy expenditure conducted during a 20-week running race across the United States.

The study is among the first to quantify the upper limits of human daily energy expenditure and endurance, whether someone is running across the country, competing in the Tour de France or pregnant. The study’s counterintuitive findings have implications for athletes, our understanding of human evolution, and our hopes that training for a marathon or other endurance event will help us shed weight.

To most of us, it seems obvious that when we are physically active, we burn more calories than when we are sedentary. The harder or longer we work out, the more of these calories we will expend, ad infinitum.

But a small but growing body of research suggests there are limits. A 2012 study of energy expenditure among modern hunter-gatherers, for instance, found that despite being in motion almost all day, the tribespeople burned about the same number of daily calories as those of us who sit behind desks all day. In effect, the tribespeople’s bodies seemed to have found ways to reduce their overall daily energy expenditure, even as they continued to move.

The study’s authors concluded that this finding made sense from an evolutionary standpoint. The fewer calories our forebears had to expend on days when they hunted, the less food they would need to bring down.

But the human caloric ceiling remained unknown and difficult to quantify. Finding it, the scientists reasoned, would require studying people who were exercising regularly at or near their physical limits and seeing how their metabolisms responded over time.

Then, in 2015, the right situation arose. For a one-time event called the Race Across USA, participants would cross the country on foot from California to Washington, D.C., running approximately a marathon almost every day for about 20 weeks.

A group of scientists, including some who had conducted the 2012 study of hunter-gatherers, asked to monitor the racers’ metabolisms. Six participants agreed, and the researchers measured their baseline daily energy expenditure in the week before they began racing. They used a gold-standard technique called doubly labeled water, in which hydrogen and oxygen are replaced with isotopes that trace the body’s production of carbon dioxide.

The researchers repeated the metabolic testing each day during the first week of daily marathons, and then again during the runners’ final week, about five months later. (Only three of the original volunteers remained in the race.)

The changes in the runners’ energy expenditures were striking. In their first week of repeated marathons, the runners burned about 6,200 calories a day on average, a steep increase over their typical energy expenditure from the week before — and about what would be expected, based on their new level of activity.

But 20 weeks later, although they were running just as much and at about the same pace, the racers had lost little body weight, and were expending about 600 fewer calories each day on average than they did in the first week.

By the end of the event, the researchers calculated that the runners were expending about two and a half times their resting metabolic rate each day, a notable decline from the early days of the event, when they were burning at least three and a half times their resting rate.

To better understand their finding’s significance, the researchers combed through the few past studies of energy expenditure using doubly labeled water. Those involved participants in other long, grueling physical endeavors, including the Tour de France bike stage race, Arctic expeditions, ultramarathons, marathons and even pregnancy.

The researchers found that in any event that lasted longer than about 12 hours, participants’ energy expenditure tended to increase substantially and then, over time, decline, until it plateaued at somewhere near 2.5 percent of their everyday, resting metabolic rate.

The researchers also examined past studies of overeating, in which people gorged on food to see how much weight they gained and how quickly, and found that most of them added pounds at a rate that suggested they could absorb about two and a half times their basic caloric needs. That is, participants might swallow more calories, but their bodies could not process anything beyond that limit.

Our bodies seem somehow to have become capable of recognizing when we are in danger of breaching the barrier beyond which we cannot easily replenish lost energy — about two and a half times our basic metabolic rate, according to Herman Pontzer, an associate professor of evolutionary anthropology at Duke University, who oversaw the new study with John Speakman and others. If we repeatedly approach that barrier, by, for instance, running marathons day after day, we apparently reduce our daily energy burn.

How our bodies manage that feat is still mysterious, Dr. Pontzer says, although the process probably involves unconsciously moving less when we are not exercising and shifting energy away from some physiological processes, such as our immune response.

For elite athletes, these findings suggest that at least part of what limits human endurance, preventing athletes and explorers from going longer and faster than they already have, is that “people can only eat so much,” Dr. Pontzer says.

The flip side of that message, though, might apply to those of us who will never run back-to-back-to-back marathons, but may hope that training for a single race will lead to weight loss, he says.

The same finely tuned physiological mechanisms that reduce energy expenditure in multiday endurance racers may slow metabolic responses over weeks and months of marathon preparation, he says.

Perhaps that explains why some people who train for a marathon gain weight.