Every year, the United States Department of Energy’s far-flung operations generate millions of tons of climate-warming greenhouse gases. By tightening valves, replacing worn gaskets and such, Josh Silverman and the department’s engineers have managed to cut the annual leaks of one gas by about 35,000 pounds.

Which would seem a pittance. Except that this gas is sulfur hexafluoride, the most potent greenhouse gas in existence — in fact, 23,900 times more potent than carbon dioxide. In three years, they have stanched leaks equivalent to 1.1 million tons of carbon dioxide.

It is as if 200,000 cars were taken off the road for a year. And that is but the beginning. “We’ve cut emissions in half,” Mr. Silverman, who heads the antileak effort as director of the department’s office of sustainability support, said this week in an interview. “The gains are real and sustained, so we’ll continue to see savings.”

It has elevated Mr. Silverman to a finalist for the Service to America Medal, one of the highest honors awarded a federal civil servant. And the accomplishment is all the more remarkable for the fact that a few years ago, nobody suspected that such a leak problem even existed.

The crusade against SF6, as sulfur hexafluoride is widely nicknamed, stemmed from an executive order early in President Obama’s first term that directed federal agencies to calculate the amount of greenhouse gases their operations generated, and then to work to reduce the totals.

Energy Department experts put paper to pencil and quickly found the equivalent of nearly four million tons of carbon dioxide, the standard measurement for greenhouse-gas emissions, in everything from power generating stations to fleets of automobiles. No one had ever systematically examined gas leaks, so they employed and then refined a simple measure: comparing the amount of gases that were bought each year with the amounts that were supposed to be in stock.

The department turned out to be a sieve. Gas leaks made up nearly one-seventh of the total 4.4 million tons of greenhouse gases they found — the equivalent of about 770,000 tons of carbon dioxide. And more than 90 percent of the leaks were of SF6.

“We were really shocked to discover the size of the releases,” Mr. Silverman said. “We had no idea of the quantity of gas that was escaping into the environment.”

The department’s string of national laboratories, many of which consume vast amounts of electricity for high-energy experiments, were the main culprits in releases of sulfur hexafluoride. SF6 is a superb electrical insulator, and laboratories employed it not only to shield high-voltage lines and equipment, but also for uses like dampening sparks in certain experiments.

The laboratories had imposed tighter controls on SF6 over the decades as its cost rose and its climate-warming potential had become better known. But the discovery that huge leaks still existed led to an extensive, and in many ways remarkable, effort to contain it.

Few were more remarkable than at the Princeton Plasma Physics Laboratory, which the university operates under contract with the Energy Department. A center for nuclear fusion research, the lab snakes with 100,000-volt cables whose electricity powers equipment that heats plasma to temperatures hundreds of times hotter than the sun’s core, and powers electromagnets that trap the heat in a so-called magnetic bottle.

The lab was losing 3,200 pounds of sulfur hexafluoride every year, the equivalent of nearly 39,000 tons of carbon dioxide.

“We said: ‘Wow. We’ve got some work to do,’ ” Robert S. Sheneman, the lab’s deputy head for environment, safety, health and security, said in an interview.

The task was daunting: more than a mile of pipes insulated with SF6, countless seals, valves, gauges and connectors, some decades old. Many were buried inside boxes or massive metal tanks, called high-voltage enclosures, that could be opened only with cranes.

And every component had to be inspected.

“Imagine your car radiator had lots of little pinholes, and you get all kinds of spray out of them,” said Tim Stevenson, the laboratory’s head of projects and a key figure in the leaks search. “At a molecular level, that’s what we were getting.

“You’re not talking about leaks that you can see or hear or even think about, or measure on a gauge and see a needle going down. This is looking for molecules of SF6 coming out in a tiny invisible plume.”

Mr. Stevenson had one stroke of luck: For much of the time, the laboratory was replacing its plasma-producing machine with a new and more powerful one, the National Spherical Torus Experiment. The switch gave engineers the time to hunt for emissions, and the opportunity to poke into equipment that was being dismantled anyway as part of the construction.

So far, engineers have cut emissions to 920 pounds from 3,200. When the new plasma machine gears up — with much more SF6 than before — that total is likely to increase a bit, Mr. Sheneman said.

But perhaps not for long. In the search for leaks, “the low fruit has already been harvested. That’s a challenge,” he said. “But it’s part of the culture here. People like a challenge.”

A version of this article appeared in print on June 14, 2013, on page A15 of the New York edition with the headline: Agency’s Crusade Against Leaks of a Potent Greenhouse Gas Yields Results