The Jet Propulsion Laboratory resembles a university campus in many ways. It has its own street names and is policed by its own security service. Shuttles take scientists and visitors around the 177-acre complex. Employees play basketball behind an outdoor sandpit where Mars rovers are tested.

From the classifieds section of the employee newsletter, called the Universe, you can purchase a mountain bike, a computer, or exercise equipment. Or rent a two-bedroom apartment.

One glaring exception is the tight security -- one gate in and out, a fence around the entire compound, and the requirement that everyone wear an identification badge.

Otherwise, the atmosphere is typical California, laid-back. A few wear ties, but some are in sandals and shorts. Employees stop to chat in the sun as they stroll from building to building up steep hills under purple-flowering jacaranda trees. By and large, they come off as a content lot. And why not? JPL is a very cool place to work.

"I come to work and I play with toys," says Bob Anderson, who helps engineer Mars rovers.

Anderson is a modern-era employee at a place that once built rockets but is now one of the world's leading robotics institutes. But Anderson is at heart a geologist, and like many scientists here, he faces a daily tug-of-war between the science he'd like to work on and the engineering he's required to do.

"It's very difficult to do your science," he says.

And it has become more difficult in the past decade. During the time that JPL was ramping up to do more missions, funding grew modestly. But the staff of full-time and contract employees was slashed from a peak of 7,608 in 1992 to just 5,098 last year. The cuts were fueled by a political push from Washington to farm projects out to private industry in an effort to broaden involvement in the space industry, Elachi said.

But the exodus reverberated through JPL. The bulk of the early departures did not involve senior-level researchers, but rather support staff: junior managers, software engineers, secretaries and the like.

"As they started dumping analysts and software engineers, there was more pressure on scientists to do their jobs," says astronomer Jim Schombert, who left JPL in 1996 for a position at the University of Oregon. The inability to do basic science was part of the reason Schombert left, though he's quick to point out that the general bureaucracy of the place took its toll, too.

"Managers would mouth that you should do good science," Schombert says. "But they never gave me the time, opportunity or resources." While glad to have moved on, he's not bitter. "This is something you buy into. You know JPL is this way going in."

If staff departures through the mid-90s were bleeding JPL of talent, the red-hot tech economy that would follow caused a hemorrhage.

Talented JPL researchers, especially technologists, became a sizzling commodity in the late 90s and into last year. Private industry sometimes offered "twice, maybe three times as much" money, Schombert says.

Opportunities went beyond just paychecks. Some JPL scientists realized life-long dreams of starting their own companies, often with seed money supplied by other major research firms.

Further, several long-time, top-level personnel reached retirement during the 1990s, one outside analyst said, and hiring cutbacks in previous years has left JPL with a shortage of qualified people ready to step into those roles.

"There has been a big brain drain," Schombert said. No one at JPL denies it.

"In terms of senior researchers, we haven't been able to replace the dozen or so people we lost over the past three years," says Paul Maker, who manages JPL's Electron Beam Lithography lab. Maker said the departures left a void of knowledge, sometimes decades of wisdom, that is not quickly filled even if replacements are hired.

But Maker says as the economy cools, there may end up being an abundance of start-up research firms that isn't supportable in the long run. "We're at the wrong end of the pendulum swing right now," he says. "But it'll come back, and we'll survive."

Naderi, the head of the Solar System Directorate, says the brain drain is a problem suffered by all high-tech organizations of late, but he admits that JPL "had a hard time keeping up" with compensation packages offered by the private sector. But, he argues, "the typical 20-year engineer didn't leave for a dot-com." He also notes that already quality applicants are starting to return.

When Elachi took over as director, one of his first efforts was to get to the bottom of the brain drain. He interviewed key employees who had left, and said their number one complaint was no surprise: The bureaucracy had left JPL less nimble and more frustrating than in the past.