(FORTUNE Magazine) – You can't stay at the top of a fast-growing company without a substantial ego. But really managing such a company requires managing your ego as well--as leaders of FORTUNE's 100 fastest-growing companies have had to learn.

Some people believe they're indispensable, but only because they haven't been severely tested. Some believe they won't fall prey to the obvious pitfalls because they've somehow avoided them in the past. And some people ignore the benefits that others bring because they suffer from a third kind of arrogance, better known as complacency.

"When executives have never had a failure, they tend to believe they're bulletproof or omnipotent," says Richard Hagberg, a psychologist whose firm, Hagberg Consulting Group, has been studying the relationship between ego and effective leadership at companies. Not surprisingly, he found that overconfidence can have "a massive negative effect." Just ask any Sunbeam shareholder about Al Dunlap's reign.

The following leaders of three of America's fastest growers learned to cope with the damage that out-of-control egos inflicted on their organizations. David Pomije founded and runs Funco (No. 2), the nation's largest retailer of used videogames. But it was only after he torpedoed two businesses that he realized he was his company's biggest problem. For Bill Morean, who took over circuitboard maker Jabil Circuit (No. 6) from his father, a string of successful ventures made him believe that he could do no wrong; in that, he was wrong. And for Jim Day, who runs offshore-driller Noble Drilling (No. 1), it took years of watching his executives hire personnel just like themselves before he found an objective method for hiring the best.

David Pomije describes himself as "a wild and crazy entrepreneur." He's got a point. By age 31, he had already started and sunk two businesses. Pomije had trusted only himself to make every important decision. His first failure, a membership-based travel club called Club Aire, lasted a year, undone by high leverage at a time of rising interest rates. The second company, Protectronics, bought Commodore 64 products at 30 cents on the dollar and resold them at a profit. For a while overruns kept the products plentiful and cheap, but when they dried up, all Pomije had left was a $300,000 bank note co-signed by his father, a high-school teacher. Pomije declared bankruptcy and set about liquidating the company.

"I felt like I had cornered the market in losing money," laughs the sandy-haired entrepreneur, leaning against a wall at his Eden Prairie, Minn., headquarters.

Amid the assortment of joysticks, software, and Commodore 64 junk to be liquidated was a library of 1,100 Nintendo games. Pomije received three bids. Just as he was about to accept the highest, his mental videoscreen lit up. Why not buy the games himself and rent them out to videogame junkies? With his dad again co-signing, Pomije persuaded the bank to lend him $8,000, the highest bid. Thus was Funco born.

Ten years later Pomije (pronounced Pom-i-JAY) oversees 261 FuncoLand stores in 22 states. Last year Funco bought and sold 9.7 million Nintendo, Sega, and Sony titles, making it the nation's largest retailer of used videogames and the largest retailer specializing in new and used ones. In the fiscal year ended in March, Funco netted $8.3 million--or $1.26 a share--on $163 million in sales. Pomije's 28% stake in Funco is today worth about $23 million.

Funco's business has changed quite a bit since Pomije dreamed it up. Originally, Pomije rented out his games from corners of local video stores, giving them half the profit. He quickly realized that renters soon grew weary of their favorite games but had a bottomless appetite for new ones. So in 1989, Pomije bought an ad in Electronic Gaming Monthly featuring 100 Nintendo titles he was looking for or wanted to get rid of and a bid and ask price on each. A week after the magazine went to subscribers, Pomije found three envelopes in his mailbox and one shoebox sitting below it. The envelopes contained checks; the shoebox held three games for sale. "I felt tingles up the back of my spine. I knew if I could do this a thousandfold, I had a business," he says. Pomije moved the operation out of his house and opened his first retail store, dubbed FuncoLand, in Minneapolis in October 1990.

But a business can't be built on tingles alone; you need sound execution to back up a great idea. Pomije used to think he was the best at every aspect of his businesses, from marketing to strategic planning, from accounting to operations. That's easy to think when everybody knows bad news is unwelcome. "My philosophy was: 'I'm the boss, you work for me,' " Pomije says. At Protectronics, employees were so scared of him they doctored inventory counts to make things look better than they were.

Today, Pomije's management style is totally different. For one thing, self-importance is out. Soft-spoken, but with a wild streak, Pomije is more likely to play a practical joke than order people around. (When a reporter was visiting, Pomije congratulated a worker on her new office and sent her off to find it. The "office" turned out to be a closet with her nameplate above the door.) And he's gotten cautious. When Funco's first two stores became profitable, Pomije paused. "I said, 'Time out. We have to stop here.' " In the old days, he would have expanded like mad. This time, although prospects were good, he focused on the cracks that he suspected lay below the surface. "I didn't want to screw up by pursuing growth for growth's sake," he says.

He didn't want to stand pat either. But if he was going to take the company to the next level, he realized he would have to modify his ingrained operating style. For six months, Pomije pored over every aspect of the business--store layouts, marketing plans, cash management. He discovered that Funco was overstaffing, and overstocking, its stores and that the marketing mix wasn't right. Then Pomije did something he had never done before: He brought in experienced people. More than anything, Pomije feels it was that that enabled Funco to grow from a 25-person startup into a competitor of Wal-Mart and Toys "R" Us. He recruited a comptroller from Wilson Leather and a systems specialist from B. Dalton. He even brought in an operating chief, a Pillsbury veteran.

And out went the trappings of power. With, for the first time, a successful business, Pomije didn't need them. Headquarters is located in a nondescript office park in a Minneapolis suburb. Employees sit in cheaply painted used cubicles costing $4 a panel. Pomije's private office isn't much better: '70s-style couches, their leather upholstery cracked and worn, occupy one corner. Across from them is a build-it-yourself entertainment center. The pieces are a symphony of mismatched Formica wood grains.

So what does Pomije do with the time his diminished need for control has freed up? With his lieutenants running the day-to-day operations, the college dropout keeps busy advising and financing other entrepreneurs. He sits on the board of a company trying to resurrect the Excelsior-Henderson motorcycle of the 1920s, and he's helping a younger version of himself start a chain of stores that sell used golf clubs.

Pomije knows he's good at startups, but less so at the later, consolidating phases of a business. That kind of self-awareness has been a blessing for Funco. Says Pomije: "I would have run Funco out of business by now if it hadn't been for the first two failures."

At Jabil Circuit, the problem wasn't a cult of personality or a reluctance to turn to others for help. It was rather the belief that the company as a whole could pull off anything it tried. For a while at least, that belief was well founded.

When Bill Morean, Jabil's chief executive, arrived in 1977, it was a barely profitable maker of circuitboards: Companies that were deluged with work would send Jabil circuits and boards, and Jabil's 90-person staff would manually assemble them as though they were Legos. Morean's father co-founded the company, then based in Detroit, as a sideline to his contracting work. But when it began to consume too much of his time, considering how little income it produced, he asked Bill, then 21 and doing construction work in Alaska, to take over.

The younger Morean's only exposure to the electronics business was working summers for Jabil during high school. That didn't stop him. He began by talking up Jabil's ability to build boards more cheaply than businesses that used them could manage in-house. In 1979, Jabil won a $12 million contract from General Motors to produce a unit allowing a car's transmission to communicate with its computerized engine control system. Every GM car needed one, amounting to three million to four million circuitboards a year, and they all had to work perfectly. That meant automating, bringing in quality-control people, and buying parts. Jabil, averaging about 6,000 circuitboards a year, had no experience in any of those areas.

So what did Morean tell GM? No problem. "That," Morean says in a slight Michigan accent, "is the benefit of being very confident and very naive at the same time." Morean quickly ramped up, hiring 120 people. Unlike Pomije, Morean knew his limitations. He brought in a confidant, Tom Sansone, then a tax lawyer with a big Detroit firm, to help lead the company. Morean also brought in a chief financial officer, a plant manager, and an engineering manager. The new team achieved flawless production of GM's parts, and in two years Jabil grew from a company grossing $400,000 a year to one grossing about $9 million.

In the mid-1980s, Morean moved Jabil's headquarters from Detroit to St. Petersburg and diversified by contracting with IBM to build add-in boards for its early personal computers. Morean and Sansone, now bearing the title of president, financed the business' growth almost exclusively with bank debt--at one point the two were personal guarantors of $80 million in loans. But the risk was strictly academic: Jabil couldn't lose. As the personal computer market boomed, so did Jabil: It made motherboards for Dell and Japanese giant NEC, circuitboards for Quantum's high-end hard drives, and controller cards that would end up in Hewlett-Packard printers. Customers could then focus on marketing and development.

Jabil went public in 1993, selling 18% of its equity for $7 a share, giving Morean a stake worth $58 million and lifting the debt off his back. But that's not the end of the story. Flush with success and the memory of every previous challenge faced and surmounted, Morean decided to move Jabil from just building computer parts to actually designing, building, and packaging the whole machine. With its partner, Epson, Jabil was going to produce the high-end Action Note 700 notebook. "We were experts at making the difficult things that go into a notebook," says Morean. "We thought, 'The rest of it is just plastics. They've been doing plastics for what--80 years? We'll get a plastics partner, and we're off to the races.' " But Morean hadn't allowed for the frequent change orders that occur during the design stage. When Jabil was about to start building, Epson decided to stop using Intel processors, delaying the launch. The hinges Jabil bought contained a lubricant that cracked the plastic. And because its experience was pretty much limited to contracts specifying a certain amount of goods for a certain price, it didn't prepare one spelling out who was responsible for what. Making the notebook "forced us into a box of technical know-how that we never intended on having," Morean says. Epson quit paying, and Jabil sued both Epson and its plastics supplier.

Four years later, Morean looks back at the Epson fiasco as a growth experience. After numerous discussions both in the boardroom and over drinks, the company set up a process--involving an assessment of its capabilities and weaknesses--for evaluating future projects. There would be no more assuming it could accomplish anything it attempted. Since then, Jabil has shot down opportunities to become a product company--by selling motherboards to the aftermarket or building its own PC--and to start a venture capital fund. Employees had been pushing such a fund as a way to lock in future customers: underwrite them, then get them to produce their circuitboards at Jabil. "No way," said Morean. "On its face, it's a pretty cool plan," he says. "But with our margins, it would be easy to give back a quarter's worth of profitability due to one startup company that leaves you holding the bag on $15 million worth of materials." As for selling motherboards, that required a whole new tool kit. "We know how to sell manufacturing strategies to our customer base," Morean says. "What we don't know is how to sell something that people are going to buy out of a Sears catalog." Instead, growth has resulted from adding to Jabil's original strengths. Where once it would send finished goods to Cisco, for example, now Jabil, at Cisco's request, sends the routers it has assembled directly to end-users. "When you're cruising along, there are lots of opportunities on the side of the road," Morean says. "You can't be distracted by every shiny object you pass."

For Noble Drilling, based in Houston, the challenge was cutting egos down to size before they hurt the company. Noble's employees spend as much as 28 days in a row surrounded by choppy waters instead of friends and family--conditions that only certain personality types can handle. The goal, as Chief Executive Jim Day sees it, is to make sure Noble hires and promotes the right ones. For six years, Day worked in the human resources department of Noble's then parent company, Noble Affiliates (No. 24). He saw firsthand that the people making personnel decisions often end up hiring and promoting people like themselves, instead of those best qualified for the task at hand. In 1983, Noble sent Day from headquarters, in Ardmore, Okla., to Tulsa to oversee Noble Drilling. The following year, Day became Noble Drilling's chief executive, and the year after that he spun it off. Today, with close to 50 rigs, it is one of the largest offshore drillers in the world.

While its success has been impressive, it hasn't always been smooth. From his first day at Noble Drilling, Day tried to figure out why the offshore operations weren't uniformly efficient and trouble-free. One rig might suffer little downtime and few accidents, while on another such problems seemed endemic. Given the dangers--whitecaps, heavy winds, and cranes that carried 10,000-pound loads swinging overhead--it was essential to get things right.

Enter Roy Rhodes, a former psychotherapist with a doctorate in education. Day and Rhodes met in 1977, when Rhodes had just embarked on a new career as a management consultant. Noble Affiliates retained Rhodes to conduct a survey of what employees saw as the company's strengths and weaknesses. That led to a program for testing all employees being considered for a promotion. In 1996, Rhodes was hired full-time and now oversees seven people in his role as vice presi- dent of organizational development. Although the title makes it sound like a make-work position for an ousted CEO, the job is taken very seriously at Noble. Before a professional or a manager is hired, he has to take four exams: two that test personality, one that tests interpersonal skills, and one that tests intelligence. To win a promotion, an employee must also be recommended by his supervisor.

Three job levels exist on offshore rigs. The entry-level floorman holds the drill pipe and puts the bit on the drill. Above him is the derrickman, who, from his perch 140 feet above the water, handles the top of the drill pipe. The top position is driller, who supervises the whole team. When a worker is about to be promoted to driller, he takes Rhodes' tests, which last about three hours, either at one of Noble's offices or on a rig. Rhodes reviews all the tests and analyzes the results.

For the past ten years, the company has collected leadership and psychological profiles revealing which traits are best suited to which positions. The exams are graded on the basis of the profiles. When an employee is under consideration for a new position, he takes the four tests again. If he fails the tests this time, a demotion is possible. The goal is to assemble teams of workers who collectively possess the cooperativeness and other traits necessary for managing a rig. "It's all standardized," brags Day.

"Most companies promote people they like," says Rhodes. "We see way too much of that. It's like incest in a herd of cattle. They're not bringing in fresh people or people who will ask a lot of questions. The oil industry is known for that."

The testing has been especially important in the past three years. Noble has been on an acquisition spree and added about 1,000 employees, boosting its total work force to about 3,500, spread over five continents. When Noble buys a company, all its employees must go through the same testing program as Noble veterans. No one, says Day, has refused to take the tests. "This isn't like school," he says. "Most people want to ascend. They're willing to step in and take the test to get higher pay and more responsibility."

While Rhodes and Day are the true fathers of the program, an obsession with talent and qualifications has long been part of Noble's culture. Posted on each rig is an excerpt from the will of founder Lloyd Noble, who died in 1950, at age 53. In it Noble declares that it's the duty of management to "weed out those who do not want to make a sincere contribution in order that room may be made for those who do."

And the size and condition of his ego? Noble built the company from a one-rig operation, financed with a $20,000 note co-signed by his mother, into a 35-drill enterprise at the time of his death. Yet in his will, the founder expressed the hope that he "would be missed personally, while the machinery continued to function smoothly." He explained: "No individual builds anything worthwhile by his efforts alone."