(FORTUNE Magazine) - Domagoj Vucic didn't come to Genentech for the rich stock options or the free cappuccino or the made-to-order sushi or the parties every Friday night. He came from the University of Georgia seven years ago because he believed Genentech could help him answer a burning question: What is it that keeps caterpillars infected with baculovirus alive for an entire seven days before they explode into a gooey puddle? Figuring that one out could, believe it or not, be a big step toward curing cancer.

Doctor-scientist Napoleone Ferrara didn't come for the perks either. He joined Genentech in 1988 because the company would allow him to pursue an obsession: the study of the formation of blood vessels that feed, say, a tumor, and the search for an antibody to disrupt the process.

It's not just the bioscientists. Ask Cynthia Wong, a mother of two, why she chose to settle at Genentech after working at Citibank and Towers Perrin, and she doesn't even mention the onsite day care or the concierge service that can pull off a birthday party on a moment's notice. Instead the senior manager quotes a breast-cancer patient who had visited her sales department a day earlier. "She has two little girls," Wong says, getting tears in her eyes all over again. "She wants to see them in braces. She wants to be there when they pick out their prom dresses." With the help of a Genentech drug called Herceptin, she probably will.

Work that really matters--it's what makes Genentech the Best Company to Work For in 2006. But there's plenty else to like about this low-key, high-tech biotech located just north of San Francisco International Airport. For starters, 29-year-old Genentech is not just the very first biotech; it's the brightest star in a promising industry that has chronically underdelivered. The company's year-end revenues should come to $6.6 billion, according to Wall Street estimates, triple what Genentech pulled in four years ago. Its stock price has doubled in the past year, to $95 a share. At one point in December it had a market cap of $102 billion, making it the 20th-most-valuable company in the U.S., ahead of Merck, Lilly, and every other pharmaceutical company except Johnson & Johnson and Pfizer. "There is only one drug company that has really come from nowhere to be a major force in this industry, and that is Genentech," says Peter Tollman, a senior vice president and biopharmaceutical expert at Boston Consulting Group. "It is the only drug company in the world that has created that much value without a merger."

Genentech's secret, anybody here will tell you, is its culture. And that is what has propelled the company to the top of this year's list. With its storybook view of San Francisco Bay, the place feels more like a college campus than a pillar of the FORTUNE 500. Signs point to the North Campus, down by the water, and the South Campus, up on the hill. Employees don't get assignments, they get "appointments." They traverse the grounds by shuttle bus and bicycles provided by the company. Every Friday night there's at least one "ho-ho"--Genentechese for kegger--a tradition that began in the '70s when the workforce was mostly a handful of rowdy postdocs barely out of grad school. At Genentech, every milestone calls for a party and a commemorative T-shirt--and on very big occasions, very big celebrity bands. A year and a half ago, after an unusual run of FDA approvals, the parking lot in front of Building 9 became the site of a rock concert featuring Elton John, Mary J. Blige, and Matchbox 20.

All this would be way too dot-com to make business sense if it weren't for another performer who took the stage that day--and who got about as much applause as the bands. That would be Art Levinson, Genentech's impish, brilliant scientist CEO, dressed for the occasion in tennis shoes and a black CLONE OR DIE T-shirt. The 55-year-old Levinson, who once bet his colleagues that five of them could fit inside an ice machine (they did), has made mostly right bets for the company ever since he took the helm in 1995--championing its science, creating a stream of new drugs, and winning over employees by making clear to all that there would be no butt-covering culture at Genentech.

In fact, Genentech's culture has a whole lot in common with those of two other Bay Area superstars: Google and Apple. All three imbue employees with idealism. Apple keeps a laserlike focus on the customer. Google's motto is DO NO EVIL. Genentech's: IN BUSINESS FOR LIFE. All three companies flout conventional wisdom and take a damn-the-torpedoes approach to naysayers. All three know one another well. Levinson is on the boards of Google and Apple; he and Steve Jobs are often seen hanging out at Genentech, deep in discussion. (Google hasn't been around long enough to qualify as a 100 Best Companies to Work for candidate; Apple declined to participate in the survey.) And all three put huge emphasis on attracting the best and the brightest. Genentech awards sabbaticals to stave off burnout. To keep creativity alive, both it and Google encourage their scientists and engineers to spend fully 20% of each workweek pursuing pet projects. Many corporations think it's terribly cutting-edge to maintain an arm's-length relationship with employees. These guys want you to move in.

Of course, it's easy to be a generous, enlightened employer when profits are flowing and your stock seems to have nowhere to go but up. IBM was that way for decades, once. But the awful mathematical truth--you don't need a Ph.D. to figure it out--is that high growth at this rate can't last forever. Says David Botstein, former head of Genentech research and now a professor of genomics at Princeton: "At some point, someone, maybe Art, maybe his successor, is going to have to figure out how to transition Genentech into a steady-state company." Levinson says he won't worry about growth slowing until sometime after 2010. Right now, with the company averaging 151 new employees a month (headcount at presstime was 9,314), he says, "The thing I worry about most is managing our growth." And protecting Genentech's mission, focus, and culture. "It's much easier to get alignment when you have fewer people."

Since the day the company was founded in 1976, Genentech's culture has been its competitive advantage. Founders Bob Swanson, a 29-year-old venture capitalist who studied the power of teams at MIT, and Herb Boyer, a pioneer gene splicer from the University of California at San Francisco, knew the success of their venture depended on luring and keeping big-brain bioscience talent. Within two years Genentech had concocted human insulin, which in 1982 became the first biotech drug to go to market. The company turned its first profit in 1979, the year before it went public, and has remained profitable ever since--despite a close call in the late 1980s, when it lost focus and stopped introducing new drugs, and the stock price flagged badly enough to make the company a takeover target. (Swiss drug giant Roche took a majority stake in 1990 and has let Genentech run independently ever since.) When Levinson was tapped to become CEO in 1995, Wall Street was skeptical but insiders cheered. Levinson was head of research, a top scientist, and astute at making calls on people and R&D.

During his first two years as CEO, Levinson persuaded the board to plow 50% of revenues back into research. (You read that number right. It's why, he believes, four of the company's 13 drugs are less than three years old, 30 more drugs are in the pipeline, and all eight of its clinical trials last year were successful.) He also decided to focus the company's science on "significant unmet needs" in the fields of oncology, immunology, and tissue growth and repair. He got rid of projects (and people) that didn't fit the program and forced fiefdoms like product development and basic research to work closely together. To head drug development, he tapped Susan Desmond-Hellmann, who had begun her career as an oncologist and has never forgotten what it's like to tell a young mother with breast cancer that she has run out of options. (Desmond-Hellmann is now president.)

Genentech pours tremendous energy into hiring people with that kind of passion. In fact, it can take five or six visits and 20 interviews to snag a job. The process is meant partly to screen out the free agents --people preoccupied with salary, title, and personal advancement. If candidates ask too many such questions, "Boom, wrong profile," says Levinson. The gantlet is also designed to let job candidates know exactly what they're getting themselves into. "We're extremely nonhierarchical," Levinson says. "We're not wearing ties. People don't call us doctor. We don't have special dining rooms." (They aren't even assigned parking spaces, and it's hell in the morning to find a spot.) Executive job seekers from Big Pharma, especially, find that a jolt, he says. "A lot of them say, 'But I like being different! I like being special!' Well, you're not going to be special here. If that's important to you, that's fine. But you won't be happy here."

Genentech looks for people who are wired like Ellen Filvaroff, a senior scientist in molecular oncology. Her walls are decorated with pictures of her patents and her toddler--side by side. The perks she likes most are little things--like being able to buy birthday cards and stamps and to mail packages from the company store. But the biggest kick by far, she says, is having colleagues who can help you crack the science more quickly. Collaboration is easy and encouraged. Once, when Filvaroff wasn't sure how to set up an experiment, she caught Napoleone Ferrara in the walkway between buildings 10 and 11. By the time they reached the doorway Filvaroff had refined her thinking enough to conduct the experiment, which turned into a published paper. Encounters like that, she says, "just bootstrap my science." As a mother of a young child, she's the least likely person to attend ho-hos. But the interplay of bioscience and brewski makes for unusually rewarding keggers, so she goes when she can.

Here status is conveyed not by snagging the fanciest title or the biggest office (CEO Levinson's measures about 9 feet by 12 feet and is done up with low-end metal office furniture). It's defined by matching wits and taking chances. Or seeing who can take the dare. At Genentech nobody dresses up, except on Halloween. This past Halloween, Desmond-Hellmann spent the day as Snow White, and Levinson and the rest of the management team dressed as the Six Dwarfs (minus Dopey). They were en route to hand out candy at another office across town when their SUV convoy drove by archrival Amgen. Levinson hailed the driver to stop and told the group he wanted to have their picture taken on the Amgen front lawn, posed around the Amgen sign. They did, but Levinson was not entirely satisfied. What he really wanted, he told them, was a picture of Snow White and the Dwarfs inside the Amgen lobby. Some of the Dwarfs chickened out, and Snow White was about to--until Levinson goaded her. "Oh come on, Sue, don't be a weenie." They entered through the revolving doors and got a shot before security guards began to arrive and they had to abort and flee. "We know the names of our patients, and a lot of them die, and I think that's part of our loopiness," explains Walter Moore, VP of government affairs.

Like Apple and Google, Genentech, despite the fun and games, is anything but relaxed. Once or twice a year, staff scientists and researchers must defend their work before the Research Review Committee, the group of 13 Ph.D.s that decides how to allot the research budget. Some find the experience nerve-wracking, and that's okay, says Levinson. "I don't want people terrified, but it should not be a cakewalk either." The rigor is designed to vet the science, uncover the flaws, avoid the dead ends, and sift out politics and favoritism, so that in the end Levinson and Desmond-Hellmann have enough information to place the right bets on the research that will most likely lead to an actual drug. Sometimes scientists fail, or the work isn't deemed sufficiently high priority, and the RRC puts an end to a project. In those cases, not only are the researchers not fired; they usually have a say in their next assignment.

True innovation takes guts. Industrywide, new drugs on average cost about $800 million and take up to 12 years to develop. More than 90% of the drugs in clinical development never reach the market, including half of those that make it to late-stage clinical trials. That's why so many big drug companies are running out of new drugs. For a long time it was easier and lucrative enough to pursue what Vishva Dixit, vice president of research, calls the "detergent" strategy--creating me-too drugs in big established markets as if they were laundry soap, and then spending big bucks on marketing to steal share from rival pharmas.

At Genentech, using market data or return-on-investment analysis to drive the science is strictly taboo. "At the end of the day, we want to make drugs that really matter," says Levinson. "That's the transcendent issue." Not that this company considers itself a philanthropy. By decade's end, it aims to be the leading U.S. oncology company in terms of sales and a leader in both immunology and tissue-growth disorders, setting ambitious new product goals in each of those categories. It has a salesforce of fewer than 1,000 and licenses with Roche and others to sell its products overseas.

Levinson really believes that if the company does the right thing, sales will follow. The strategy: Fund enough basic research in targeted areas of interest,0 and the results will yield multiple drugs--or drugs that can be used in multiple ways. That makes Genentech an especially rewarding place to work for a scientist like Ferrara, whose 17-year obsession--launched with a breakthrough made on discretionary time--led him to discover VEGF, a key to blood-vessel formation, which in turn enabled Genentech to develop an antibody that can choke off the blood supply to certain tumors. Those discoveries laid the groundwork for two of Genentech's newest drugs, Avastin, approved to treat colorectal cancer, and Lucentis, which is awaiting FDA approval for treatment of age-related blindness. Avastin might have died a premature death when it failed clinical trials three years ago, causing the company's share price to plunge nearly 10% overnight (to a split-adjusted $14.45 a share). But because Levinson and his lieutenants were so deep into the science, they knew better than to give up. Avastin, approved in February 2004, had sales of $774 million in the first nine months of 2005.

Ferrara is studying VEGF alternatives for regulating blood supply that could lead to still more new drugs. And Domagoj Vucic's work that began with caterpillars has opened an area of anticancer research so promising that an executive calls it the company's "search for the Holy Grail." It involves an effort to regulate apoptosis, the natural ability of cells to self-destruct when they've lived out their lives or are stricken with disease. For Vucic the quest is personal: His friend and mentor at the University of Georgia, who helped launch the research, died of melanoma not long after Vucic left there. It is the kind of long-term, high-risk research that makes Genentech employees proud--and glad, they say, that they aren't at Big Pharma.

Scoffing at Big Pharma may be both great sport and an effective rallying cry, but everybody here is painfully aware that Genentech also runs the risk of getting too large. By the end of this year, 40% of the workforce will have spent less than three years at Genentech; another 40% of its managers will be new to their positions. So the company is working furiously to acculturate the rookies. New-hire orientation includes patient lectures, history lessons by Boyer and other old-timers, in-depth sessions on the company's goals, its science--and the fact that the place works "because of all the thousands of little decisions that are made every day," says HR vice president Denise Smith-Hams. The company polls its workers weekly to ferret out complaints and monitor whether all the new parties are aligned with Genentech's goals.

When Levinson sees signs of culture atrophy, he pounces, as he did in an e-mail to senior managers in December about "the spread of unintelligible, gibberish-laden PowerPoint presentations.... I have recently sat through several presentations that were simply incomprehensible-- mind-numbing, bloated discourses that were full of buzzwords and otherwise devoid of meaningful content. This is a serious problem, and the worst part is that it's spreading like the disease it is." (His abhorrence of corporatespeak helps explain why Levinson loathes consultants. "They suck you dry," he says.)

In case the memo alone doesn't do the trick, Levinson invented a game called gBuzz Bingo. Here's how to play: From the company intranet, download a bingo card featuring terms like "actionable," "traction," "value-added," and "win-win." Take the card to any meeting where you expect the worst. Check off boxes as the words are uttered. First to complete a line wins, which of course requires that you shout out: "gBuzz!" The winner receives the smug satisfaction of silencing the b.s. And DNA by the Bay, as Genentechers call their company, keeps its magic--for one more meeting, at least.

FEEDBACK bmorris@fortunemail.com

Domagoj Vucic didn't come to Genentech for the rich stock options or the free cappuccino or the made-to-order sushi or the parties every Friday night. He came from the University of Georgia seven years ago because he believed Genentech could help him answer a burning question: What is it that keeps caterpillars infected with baculovirus alive for an entire seven days before they explode into a gooey puddle? Figuring that one out could, believe it or not, be a big step toward curing cancer.

Doctor-scientist Napoleone Ferrara didn't come for the perks either. He joined Genentech in 1988 because the company would allow him to pursue an obsession: the study of the formation of blood vessels that feed, say, a tumor, and the search for an antibody to disrupt the process.

It's not just the bioscientists. Ask Cynthia Wong, a mother of two, why she chose to settle at Genentech after working at Citibank and Towers Perrin, and she doesn't even mention the onsite day care or the concierge service that can pull off a birthday party on a moment's notice. Instead the senior manager quotes a breast-cancer patient who had visited her sales department a day earlier. "She has two little girls," Wong says, getting tears in her eyes all over again. "She wants to see them in braces. She wants to be there when they pick out their prom dresses." With the help of a Genentech drug called Herceptin, she probably will.

Work that really matters--it's what makes Genentech the Best Company to Work For in 2006. But there's plenty else to like about this low-key, high-tech biotech located just north of San Francisco International Airport. For starters, 29-year-old Genentech is not just the very first biotech; it's the brightest star in a promising industry that has chronically underdelivered. The company's year-end revenues should come to $6.6 billion, according to Wall Street estimates, triple what Genentech pulled in four years ago. Its stock price has doubled in the past year, to $95 a share. At one point in December it had a market cap of $102 billion, making it the 20th-most-valuable company in the U.S., ahead of Merck, Lilly, and every other pharmaceutical company except Johnson & Johnson and Pfizer. "There is only one drug company that has really come from nowhere to be a major force in this industry, and that is Genentech," says Peter Tollman, a senior vice president and biopharmaceutical expert at Boston Consulting Group. "It is the only drug company in the world that has created that much value without a merger."

Genentech's secret, anybody here will tell you, is its culture. And that is what has propelled the company to the top of this year's list. With its storybook view of San Francisco Bay, the place feels more like a college campus than a pillar of the FORTUNE 500. Signs point to the North Campus, down by the water, and the South Campus, up on the hill. Employees don't get assignments, they get "appointments." They traverse the grounds by shuttle bus and bicycles provided by the company. Every Friday night there's at least one "ho-ho"--Genentechese for kegger--a tradition that began in the '70s when the workforce was mostly a handful of rowdy postdocs barely out of grad school. At Genentech, every milestone calls for a party and a commemorative T-shirt--and on very big occasions, very big celebrity bands. A year and a half ago, after an unusual run of FDA approvals, the parking lot in front of Building 9 became the site of a rock concert featuring Elton John, Mary J. Blige, and Matchbox 20.

All this would be way too dot-com to make business sense if it weren't for another performer who took the stage that day--and who got about as much applause as the bands. That would be Art Levinson, Genentech's impish, brilliant scientist CEO, dressed for the occasion in tennis shoes and a black CLONE OR DIE T-shirt. The 55-year-old Levinson, who once bet his colleagues that five of them could fit inside an ice machine (they did), has made mostly right bets for the company ever since he took the helm in 1995--championing its science, creating a stream of new drugs, and winning over employees by making clear to all that there would be no butt-covering culture at Genentech.

In fact, Genentech's culture has a whole lot in common with those of two other Bay Area superstars: Google and Apple. All three imbue employees with idealism. Apple keeps a laserlike focus on the customer. Google's motto is DO NO EVIL. Genentech's: IN BUSINESS FOR LIFE. All three companies flout conventional wisdom and take a damn-the-torpedoes approach to naysayers. All three know one another well. Levinson is on the boards of Google and Apple; he and Steve Jobs are often seen hanging out at Genentech, deep in discussion. (Google hasn't been around long enough to qualify as a 100 Best Companies to Work for candidate; Apple declined to participate in the survey.) And all three put huge emphasis on attracting the best and the brightest. Genentech awards sabbaticals to stave off burnout. To keep creativity alive, both it and Google encourage their scientists and engineers to spend fully 20% of each workweek pursuing pet projects. Many corporations think it's terribly cutting-edge to maintain an arm's-length relationship with employees. These guys want you to move in.

Of course, it's easy to be a generous, enlightened employer when profits are flowing and your stock seems to have nowhere to go but up. IBM was that way for decades, once. But the awful mathematical truth--you don't need a Ph.D. to figure it out--is that high growth at this rate can't last forever. Says David Botstein, former head of Genentech research and now a professor of genomics at Princeton: "At some point, someone, maybe Art, maybe his successor, is going to have to figure out how to transition Genentech into a steady-state company." Levinson says he won't worry about growth slowing until sometime after 2010. Right now, with the company averaging 151 new employees a month (headcount at presstime was 9,314), he says, "The thing I worry about most is managing our growth." And protecting Genentech's mission, focus, and culture. "It's much easier to get alignment when you have fewer people."

Since the day the company was founded in 1976, Genentech's culture has been its competitive advantage. Founders Bob Swanson, a 29-year-old venture capitalist who studied the power of teams at MIT, and Herb Boyer, a pioneer gene splicer from the University of California at San Francisco, knew the success of their venture depended on luring and keeping big-brain bioscience talent. Within two years Genentech had concocted human insulin, which in 1982 became the first biotech drug to go to market. The company turned its first profit in 1979, the year before it went public, and has remained profitable ever since--despite a close call in the late 1980s, when it lost focus and stopped introducing new drugs, and the stock price flagged badly enough to make the company a takeover target. (Swiss drug giant Roche took a majority stake in 1990 and has let Genentech run independently ever since.) When Levinson was tapped to become CEO in 1995, Wall Street was skeptical but insiders cheered. Levinson was head of research, a top scientist, and astute at making calls on people and R&D.

During his first two years as CEO, Levinson persuaded the board to plow 50% of revenues back into research. (You read that number right. It's why, he believes, four of the company's 13 drugs are less than three years old, 30 more drugs are in the pipeline, and all eight of its clinical trials last year were successful.) He also decided to focus the company's science on "significant unmet needs" in the fields of oncology, immunology, and tissue growth and repair. He got rid of projects (and people) that didn't fit the program and forced fiefdoms like product development and basic research to work closely together. To head drug development, he tapped Susan Desmond-Hellmann, who had begun her career as an oncologist and has never forgotten what it's like to tell a young mother with breast cancer that she has run out of options. (Desmond-Hellmann is now president.)

Genentech pours tremendous energy into hiring people with that kind of passion. In fact, it can take five or six visits and 20 interviews to snag a job. The process is meant partly to screen out the free agents --people preoccupied with salary, title, and personal advancement. If candidates ask too many such questions, "Boom, wrong profile," says Levinson. The gantlet is also designed to let job candidates know exactly what they're getting themselves into. "We're extremely nonhierarchical," Levinson says. "We're not wearing ties. People don't call us doctor. We don't have special dining rooms." (They aren't even assigned parking spaces, and it's hell in the morning to find a spot.) Executive job seekers from Big Pharma, especially, find that a jolt, he says. "A lot of them say, 'But I like being different! I like being special!' Well, you're not going to be special here. If that's important to you, that's fine. But you won't be happy here."

Genentech looks for people who are wired like Ellen Filvaroff, a senior scientist in molecular oncology. Her walls are decorated with pictures of her patents and her toddler--side by side. The perks she likes most are little things--like being able to buy birthday cards and stamps and to mail packages from the company store. But the biggest kick by far, she says, is having colleagues who can help you crack the science more quickly. Collaboration is easy and encouraged. Once, when Filvaroff wasn't sure how to set up an experiment, she caught Napoleone Ferrara in the walkway between buildings 10 and 11. By the time they reached the doorway Filvaroff had refined her thinking enough to conduct the experiment, which turned into a published paper. Encounters like that, she says, "just bootstrap my science." As a mother of a young child, she's the least likely person to attend ho-hos. But the interplay of bioscience and brewski makes for unusually rewarding keggers, so she goes when she can.

Here status is conveyed not by snagging the fanciest title or the biggest office (CEO Levinson's measures about 9 feet by 12 feet and is done up with low-end metal office furniture). It's defined by matching wits and taking chances. Or seeing who can take the dare. At Genentech nobody dresses up, except on Halloween. This past Halloween, Desmond-Hellmann spent the day as Snow White, and Levinson and the rest of the management team dressed as the Six Dwarfs (minus Dopey). They were en route to hand out candy at another office across town when their SUV convoy drove by archrival Amgen. Levinson hailed the driver to stop and told the group he wanted to have their picture taken on the Amgen front lawn, posed around the Amgen sign. They did, but Levinson was not entirely satisfied. What he really wanted, he told them, was a picture of Snow White and the Dwarfs inside the Amgen lobby. Some of the Dwarfs chickened out, and Snow White was about to--until Levinson goaded her. "Oh come on, Sue, don't be a weenie." They entered through the revolving doors and got a shot before security guards began to arrive and they had to abort and flee. "We know the names of our patients, and a lot of them die, and I think that's part of our loopiness," explains Walter Moore, VP of government affairs.

Like Apple and Google, Genentech, despite the fun and games, is anything but relaxed. Once or twice a year, staff scientists and researchers must defend their work before the Research Review Committee, the group of 13 Ph.D.s that decides how to allot the research budget. Some find the experience nerve-wracking, and that's okay, says Levinson. "I don't want people terrified, but it should not be a cakewalk either." The rigor is designed to vet the science, uncover the flaws, avoid the dead ends, and sift out politics and favoritism, so that in the end Levinson and Desmond-Hellmann have enough information to place the right bets on the research that will most likely lead to an actual drug. Sometimes scientists fail, or the work isn't deemed sufficiently high priority, and the RRC puts an end to a project. In those cases, not only are the researchers not fired; they usually have a say in their next assignment.

True innovation takes guts. Industrywide, new drugs on average cost about $800 million and take up to 12 years to develop. More than 90% of the drugs in clinical development never reach the market, including half of those that make it to late-stage clinical trials. That's why so many big drug companies are running out of new drugs. For a long time it was easier and lucrative enough to pursue what Vishva Dixit, vice president of research, calls the "detergent" strategy--creating me-too drugs in big established markets as if they were laundry soap, and then spending big bucks on marketing to steal share from rival pharmas.

At Genentech, using market data or return-on-investment analysis to drive the science is strictly taboo. "At the end of the day, we want to make drugs that really matter," says Levinson. "That's the transcendent issue." Not that this company considers itself a philanthropy. By decade's end, it aims to be the leading U.S. oncology company in terms of sales and a leader in both immunology and tissue-growth disorders, setting ambitious new product goals in each of those categories. It has a salesforce of fewer than 1,000 and licenses with Roche and others to sell its products overseas.

Levinson really believes that if the company does the right thing, sales will follow. The strategy: Fund enough basic research in targeted areas of interest,0 and the results will yield multiple drugs--or drugs that can be used in multiple ways. That makes Genentech an especially rewarding place to work for a scientist like Ferrara, whose 17-year obsession--launched with a breakthrough made on discretionary time--led him to discover VEGF, a key to blood-vessel formation, which in turn enabled Genentech to develop an antibody that can choke off the blood supply to certain tumors. Those discoveries laid the groundwork for two of Genentech's newest drugs, Avastin, approved to treat colorectal cancer, and Lucentis, which is awaiting FDA approval for treatment of age-related blindness. Avastin might have died a premature death when it failed clinical trials three years ago, causing the company's share price to plunge nearly 10% overnight (to a split-adjusted $14.45 a share). But because Levinson and his lieutenants were so deep into the science, they knew better than to give up. Avastin, approved in February 2004, had sales of $774 million in the first nine months of 2005.

Ferrara is studying VEGF alternatives for regulating blood supply that could lead to still more new drugs. And Domagoj Vucic's work that began with caterpillars has opened an area of anticancer research so promising that an executive calls it the company's "search for the Holy Grail." It involves an effort to regulate apoptosis, the natural ability of cells to self-destruct when they've lived out their lives or are stricken with disease. For Vucic the quest is personal: His friend and mentor at the University of Georgia, who helped launch the research, died of melanoma not long after Vucic left there. It is the kind of long-term, high-risk research that makes Genentech employees proud--and glad, they say, that they aren't at Big Pharma.

Scoffing at Big Pharma may be both great sport and an effective rallying cry, but everybody here is painfully aware that Genentech also runs the risk of getting too large. By the end of this year, 40% of the workforce will have spent less than three years at Genentech; another 40% of its managers will be new to their positions. So the company is working furiously to acculturate the rookies. New-hire orientation includes patient lectures, history lessons by Boyer and other old-timers, in-depth sessions on the company's goals, its science--and the fact that the place works "because of all the thousands of little decisions that are made every day," says HR vice president Denise Smith-Hams. The company polls its workers weekly to ferret out complaints and monitor whether all the new parties are aligned with Genentech's goals.

When Levinson sees signs of culture atrophy, he pounces, as he did in an e-mail to senior managers in December about "the spread of unintelligible, gibberish-laden PowerPoint presentations.... I have recently sat through several presentations that were simply incomprehensible-- mind-numbing, bloated discourses that were full of buzzwords and otherwise devoid of meaningful content. This is a serious problem, and the worst part is that it's spreading like the disease it is." (His abhorrence of corporatespeak helps explain why Levinson loathes consultants. "They suck you dry," he says.)

In case the memo alone doesn't do the trick, Levinson invented a game called gBuzz Bingo. Here's how to play: From the company intranet, download a bingo card featuring terms like "actionable," "traction," "value-added," and "win-win." Take the card to any meeting where you expect the worst. Check off boxes as the words are uttered. First to complete a line wins, which of course requires that you shout out: "gBuzz!" The winner receives the smug satisfaction of silencing the b.s. And DNA by the Bay, as Genentechers call their company, keeps its magic--for one more meeting, at least.

FEEDBACK bmorris@fortunemail.com