(FORTUNE Magazine) – A clutch of Pentagon officials sat around a conference table on the 30th floor of Pfizer's Manhattan headquarters last month, firing questions at David Shedlarz, the CFO. They had come seeking strategic advice--in recent years the Department of Defense has been quietly calling on some of the nation's largest companies to glean ideas on how to streamline the Pentagon. The officials visiting Shedlarz wanted him to explain how the world's largest drug company has managed to grow and prosper despite its size.

They picked the right guy to debrief. Since his company announced its acquisition of Pharmacia last July--a $60 billion deal--Shedlarz has spent his days and nights orchestrating the combination. Issues of magnitude are constantly on his mind: how to integrate the companies' sprawling administrative systems; how to combine their manufacturing and distribution infrastructures around the world; how to meld Pharmacia's 4,000-person sales force with Pfizer's 8,000-person force; and on and on. At his meeting with the Defense Department officials, Shedlarz delivered a presentation packed with guiding principles like "promote delegation," "empower all workers," and "obliterate vertical hierarchies."

But the Pentagon's broadest query was one he couldn't fully answer: "How do we avoid the dark side of scale?" It's a question all great corporations, from General Motors and IBM to Wal-Mart and Citigroup, ultimately confront--how a mammoth organization can avoid the pitfalls of size, remain effective, and respond nimbly to opportunity. Or to use Lou Gerstner's memorable metaphor for IBM: Can the elephant be made to dance? For Pfizer at the moment, the question is unanswerable, and yet it guides the company's every action. Explains Shedlarz: "This is a bold, new experiment."

There has never been a drug company as big as Pfizer. It shows up on the FORTUNE 500 as a $35-billion-a-year business, but once the merger is completed later this month, it will have a projected annual revenue of $48 billion. That will make it 50% larger than its nearest competitor, GlaxoSmithKline. (Merck, whose sales appear as $51 billion on the list, consolidates $30 billion of revenues from its Medco prescription-benefit subsidiary.) Pfizer's global sales force of 35,000 will dwarf that of its nearest rival by some 7,000. And with a combined R&D budget of $7 billion a year, Pfizer will be the world's largest privately funded biomedical research operation, with no fewer than 120 "new chemical entities"--innovative drugs--in clinical development.

Even by elephant standards, Pfizer leaves large tracks. Its market cap--$178 billion as of March 14--puts it near the pinnacle of the FORTUNE 500; only Microsoft, General Electric, Exxon Mobil, and Wal-Mart have higher valuations. Pfizer's business is built on a massive foundation of nine blockbuster drugs--Celebrex, Diflucan, Lipitor, Neurontin, Norvasc, Viagra, Zithromax, Zoloft, and Zyrtec. Each is a leader in its therapeutic category, each generates sales of more than $1 billion a year, and each is growing at double-digit annual rates. Pfizer medicines dominate in three major areas--cardiovascular, central nervous system, and inflammatory-disease problems--and they touch millions of lives; you probably have at least one Pfizer prescription in your medicine chest. For most of these blockbusters, which represent 85% of Pfizer's drug revenues, the company's patents will remain in force throughout this decade.

But Pfizer's added heft has yet to impress investors. Even though its profits last year were drug-baron rich, rising 17% to $9.1 billion on $35.2 billion in sales, Pfizer's shares tumbled 23%, to around $30. That sickening decline matched the fall of the S&P pharmaceuticals index, which sank 22%.

The reasons for investors' lack of confidence? For one thing, before the Pharmacia deal, nothing much was coming out of Pfizer's new-drug pipeline. Revenue growth in 2002 was flat, compared with the industry average of 8%. For another thing, as a bellwether stock held by 2.2 million shareholders, Pfizer reflects an industrywide malaise. The problems pharma companies face are widely known, and the bigger Pfizer becomes, the more subject it is to these pressures: rising competition from generic-drug makers, increasing threats from Washington to hold down prices, and, most puzzling for an industry supposedly in the vanguard of 21st-century science, woeful underachievement in R&D. Indeed, drug research is quite possibly the least efficient endeavor in the world of business. It's the equivalent of hiring thousands of art students and funding decades of work in hopes that once in a while one will paint a "Mona Lisa." Overall, some 96% of pharmaceutical research projects fail.

Nowhere is the R&D failure rate more pronounced than at Pfizer. It outspends even its largest rival by more than 30% yet is only in the middle of the pack in R&D efficiency (see table). Between 1996 and 2001, Pfizer researchers patented 1,217 new compounds at an average cost per patent of $17.5 million. By comparison, Merck--with the industry's most efficient research arm--registered 1,933 compounds, spending roughly $6 million per patent. While newly patented compounds don't always lead to marketable drugs, they are a good indicator of research productivity. Morgan Stanley analyst Jami Rubin says Pfizer's mediocre performance "underscores the significant hurdle it faces generating a sufficient number of blockbusters out of its $7 billion pipeline. If this feat isn't accomplished, we think Pfizer will likely hit a wall and be forced into another large merger." No wonder investors have doubts.

Chairman and CEO Hank McKinnell presides over Pfizer from an immaculate corner office looking down on 42nd Street, midtown Manhattan, the East River, Brooklyn, and beyond. In his 32 years with Pfizer, McKinnell, 60, has held almost every important executive-level post, including president of Pfizer Asia, chief operating officer, chief financial officer, and president of the company's medical technology group. Warm and avuncular one minute, exacting and impatient the next, McKinnell is an odd cross between Mr. Rogers and the cranky Dr. "Bones" McCoy from Star Trek. In a single conversation he can deliver a stirring monologue about curing the world's ills, then launch into a stern rant against consumer activists who think drug companies charge too much for their wares.

McKinnell is one of the primary architects of the acquisition strategy that has supercharged Pfizer's growth. So it's startling to hear him argue that the company's acquisition days are over. Despite the fact that Pfizer commands only 11% of the global pharmaceutical market, its growth, he declares, now lies in R&D, not M&A. "This company's future will be quite different," McKinnell says. "People tend to evaluate research looking out the back window, not the front. And the '90s were really unusual."

Indeed, the '90s were a golden era for the drug industry. Between 1980 and 1990, U.S. pharma companies boosted annual R&D spending from $1.6 billion to $6.8 billion. Those massive investments paid off in a raft of major drugs: statins like Mevacor to combat high cholesterol, SSRIs like Prozac for depression and anxiety, "proton pump inhibitors" like Prilosec for ulcers, and many more.

Pfizer reaped big dividends in that period from its R&D. In 1992 it earned a reputation as the industry's most dynamic marketer by making blockbusters out of three products from its labs: the antidepressant Zoloft, the antibiotic Zithromax, and the antihypertension drug Norvasc. Company officials referred to them simply as the "triumvirate." In five years they helped Pfizer more than double its net income, to $2.2 billion on $12.5 billion of sales in 1997.

That same year Pfizer negotiated a license from Warner Lambert to co-market Lipitor, the cholesterol-lowering drug that today is the world's leading blockbuster (see "The $10 Billion Drug" on fortune.com). Then-CEO Bill Steere announced an audacious five-year goal: to make Pfizer, already the fifth-largest drug company, No. 1 in the industry by 2001.

Then came Viagra. It probably wouldn't have taken an ace marketing team and thousands of salespeople to make the first effective erectile-dysfunction drug one of the world's most prescribed medicines in 1998, but it didn't hurt that Pfizer was firing on all cylinders for Viagra's introduction. Within weeks of the drug's release, physicians were writing 800,000 Viagra prescriptions per month.

Despite the company's thumping success, Steere and his lieutenants (including McKinnell, then COO) knew they had a problem. After Viagra the pipeline was virtually dry--the labs had nothing to sustain Pfizer's rapid climb. So Steere and crew decided to leverage the company's sales and marketing strength. They made an art of locating already assembled masterpieces at other companies' labs. Then they either licensed the right to sell the new drug (as they did with Eisai Co.'s Alzheimer's drug, Aricept) or simply bought its owner. In 2000 that strategy led to Pfizer's record-setting $90 billion acquisition of Lipitor maker Warner Lambert after a brutal bidding war against American Home Products. It also led to last year's Pharmacia deal.

All the while the R&D productivity problem lurked. To hear McKinnell tell it, Pfizer fell prey to a host of factors that conspired during the 1990s to pull the industry into its recent unproductive funk. Early in the decade the tools of drug discovery fundamentally changed. The trial-and-error laboratory methods that had produced the great blockbusters were eclipsed by computer-assisted research. It enabled scientists to pursue new, more difficult mechanisms for attacking disease, such as treatments for depression and Alzheimer's that would work by affecting so-called GABA receptors in the brain. But the immediate result was a lower success rate.

At the same time the federal government became a headache. Clinton-era efforts at health-care reform threatened to constrain profits. In response Pfizer, like many companies, focused its development efforts more and more exclusively on drugs that promised a big payoff, hoping to make up in volume what they lost in profit margins. The failure rates on projects during that period soared. Says McKinnell: "We were trying for home runs, hitting some but not many." On top of that he blames Jane Henney, who became FDA commissioner in 1998, for "raising the regulatory hurdles quite significantly." Many drugs hit the FDA only to be sent back for further study and testing, "which took more time and cost us all more money," says McKinnell. In 1996, at the peak of the blockbuster wave, the FDA approved 56 new drugs; in 2000 it approved only 19. The rejection rate went up, McKinnell says, but "it wasn't that we got dumber."

McKinnell is determined that Pfizer make its labs productive again. "I don't think you can succeed as a rainmaker unless you make it rain, and I don't think you can succeed as a pharmaceutical company unless you discover and develop drugs," he says. To tackle the issue, McKinnell authorized a multimillion-dollar task force to study why so many research projects bomb. The goal of the "attrition task force" is to reduce the labs' failure rate from 96% to 92%. That sounds modest until you think about the math: What Pfizer really has in mind is to double its R&D success rate, from 4% to 8%.

The starting point of the project is what's already known about drug failures: Historically most new compounds flunk their clinical trials because they prove to be toxic, either by themselves or in the presence of other drugs. Pfizer's committee, which consisted of no fewer than 50 people when it started work in 2001, fanned out across the company, conducting thorough postmortems on past losers, interviewing the scientists, and amassing the information in a database. The data are being used to help select new lab technology and develop early screens to weed out drug candidates with undesirable traits. Research managers are also using the data to balance Pfizer's portfolio between drugs that focus on known methods of attacking diseases and drugs aimed at unprecedented ones. One direct result of the task force: In its labs in Ann Arbor, Mich., Pfizer recently hired a senior research scientist to develop tests to help colleagues choose drug candidates with better absorption and less toxicity.

If McKinnell's attrition task force succeeds, its insights could clear the way for a resurgence of creativity and entrepreneurship in the labs. If not, it could represent the kind of creeping bureaucracy that can choke off innovation--exactly what the Pentagon planners referred to as the dark side of scale.

John LaMattina, vice president of global R&D, isn't worried. He thinks bigness will only play to Pfizer's advantage. "For years we were eating Merck's dust. We'd know from intelligence that they had double the number of people on research projects. Now that our discovery group is bigger than Merck's, it's not in vogue anymore to be big. You've got to be kidding me!" he laughs. For LaMattina, who joined Pfizer's Groton, Conn., labs in 1977, being big means not having to make hard choices. "In the old days," he says, recalling Pfizer's efforts to develop cardiovascular drugs, "we had to choose between going after ACE inhibitors or sodium and calcium channel blockers. We invested a lot in calcium channel blockers, and Norvasc is a wonderful drug, but it would have been nice to go after ACE inhibitors too, and have an Enalapril," he says, referring to a Merck blockbuster. "Now we go after what we think are good targets."

McKinnell, for his part, is betting that if Pfizer can get its R&D house in order, biotech will do the rest. What about the notion that genomics means curtains for the blockbuster era? The argument is that biotech is best suited for creating drugs that treat ailments in small, narrowly defined populations of patients--for example, a drug to treat breast cancers only in women who have the BRCA1 gene. Pfizer says, No way. Rather than use biotech to craft such designer drugs, McKinnell explains, Pfizer will employ it to keep new blockbusters from getting derailed in clinical trials. Trial drugs often fail not because they are toxic to all patients, but because a handful experience adverse effects. Genomics can help researchers and doctors screen out such patients so that the majority can benefit from the new drug. "It's not as if we'll be discovering smaller drugs," says McKinnell. "The markets are still big. The drugs will just be more specifically targeted." This new breed of blockbuster, he says, will dominate the landscape for years.

Wall Street is showing signs of coming around. A good number of financial analysts believe that Pfizer's shares, hovering at $30, are trading at a discount. Says Scott Kay of Banc of America Securities: "In the '80s they were criticized for their spending; then in the '90s they yielded all these products. I'm looking at the latter part of this decade, and I'm not yet willing to say that they're guilty of investing too much money without yield."

Pfizer also has a strong near-term pipeline. Promising candidates include Pregabalin, a medicine for anxiety and epilepsy that is in Phase III clinical trials. Another promising drug, in Phase II trials, is known simply as "CP-529, 414." It has been shown to increase good cholesterol while decreasing the bad kind, and is designed to work in conjunction with Lipitor as well as when taken alone. Pharmacia brings potential blockbusters too: the cardiovascular drug Inspra, an injectable pain reliever called Parecoxib, and a rheumatoid arthritis drug known as CDP-870. What's more, completing the Pharmacia acquisition ought to give Pfizer a shot in the arm. Shedlarz says Pfizer will realize some $2.5 billion in cost savings over the next four years. And Pharmacia's R&D operations are more efficient. All told, Pfizer management is promising to outpace the industry with double-digit sales growth through 2004.

No one looks forward to continued expansion more than Pfizer's rank and file. So conscious are employees of growth that in talking about their jobs, they sometimes measure the passage of time not in years but by Pfizer's rank in the industry. At the company's 56-year-old factory in Brooklyn, you hear things like "When we were five, we introduced Viagra" and "I remember when we were ten."

But if you walk the plant with production manager Chris Mosely, you don't need to talk about growth at all. It's right in front of you. "You'll have to excuse all this," he says, staring at an eight-foot-high skid of pill-label boxes in the middle of the hallway. "We're a little limited in our space." A 23-year veteran, Mosely helps set the production schedule at the eight-story plant, one of 30 factories Pfizer owns worldwide. "It's not designed for what we do, but we've adapted," Moseley says. In the past ten years the factory has more than doubled its output. It now employs 1,200 and runs 24/7, making everything from Zithromax to Zoloft--more than two billion pills a year. McKinnell's challenge will be to make every Pfizer plant as busy as this one. If he succeeds, investors may well conclude that bigness is worth betting on.