(FORTUNE Magazine) - IS Hank McKinnell a failure? A success? Or something in between? Those are the questions people in and out of Pfizer are pondering as its CEO enters his final years on the job.

Everyone certainly agrees that McKinnell has made some noise. Earlier this decade he pulled off two of the largest mergers in Big Pharma history, creating a colossus with $51 billion in sales. Pfizer, now No. 31 on the FORTUNE 500, is the world's largest drug company. Its single bestselling drug, the cholesterol-lowering pill Lipitor, generates annual revenue of $12.9 billion--more than companies like Oracle and Marriott.

That said, Pfizer's (Research) vaunted research labs haven't discovered a genuine blockbuster since giving birth to Viagra in 1998. Profits? Down almost 30% last year. And Pfizer's shares have lost about 40% of their value since McKinnell took over as CEO in 2001, vs. a 19% drop for pharma as a whole, due to an industrywide R&D drought. McKinnell raked in some $61 million in salary and bonuses in that period, during which the S&P 500 index rose 5%.

If that isn't enough to raise questions about his legacy, McKinnell has opted to manage his succession by pitting three sharp-elbowed executives against one another, which has raised a few eyebrows within the company and on Wall Street.

Still, as far as McKinnell is concerned, you can't begin to judge him until you understand what is known at Pfizer as "the cliff." Go back to the mid-1990s, as the company was in the process of launching a dozen groundbreaking medicines like Viagra and the antidepressant Zoloft, when a confidential memo laid out a disquieting vision of what could happen sometime after the millennium.

Key patents were due to expire, and there were far too few potential blockbuster drugs in the early-stage pipeline--Pfizer was approaching a period of steep decline. For everyone privy to the report, including McKinnell, then a VP, a single page stood out. On it was a graphic that forecast a whopping 33% drop in Pfizer's revenues over a three- to four-year period starting around 2003. "We called it the cliff,"says McKinnell. "Everything I've done has been to manage that challenge, and I think we've done a pretty good job."

But as the 63-year-old CEO enters his sixth year as boss--with a couple more to go--the abyss still looms. And it's somewhat deeper. Patents on five of the company's bestselling products will expire in the next five years. This summer Zoloft, which generates annual sales of $3.3 billion, will face off against generics. In 2007 the company will lose exclusive rights to its $4.7 billion hypertension treatment, Norvasc, and its $1.7 billion allergy medicine, Zyrtec. Between 2005 and 2008, Pfizer will face $12 billion in annual revenue losses as a result of patent expirations.

McKinnell has to worry about patent expirations outside Pfizer too. Over the next few months Lipitor sales are likely to be challenged, as competing drugs Zocor (from Merck) and Pravachol (Bristol-Myers Squibb) go off patent and less expensive cholesterol-lowering generics flood the market. Making matters worse, 2005 revenues declined by 3%, largely due to fallout after Merck's 2004 withdrawal of Vioxx over heart-attack concerns. Pfizer took one of its own COX-2 inhibitors, Bextra, off the market, while another, Celebrex, remained but saw sales drop. In all, Pfizer unexpectedly lost billions in sales during 2005. (Total revenues this year are expected to be flat.)

Having skirted the abyss so far, McKinnell has a plan to return the company to growth next year. On his to-do list: revive Pfizer's research labs, now in a fallow period; execute flawless product launches while making deep cuts in sales, marketing, and manufacturing; and buy or license promising drugs from smaller companies.

He's also considering whether to sell or spin off the consumer products unit, which includes drugstore brands like Listerine and Bengay. (A sale could fetch $13 billion--and allow Pfizer to focus on prescription medications.) All this and get Wall Street to buy into the notion that Big Pfizer can keep growing.

A sharp-minded and sometimes imperious leader, McKinnell is not averse to taking the odd risk. As a teen he was a deck hand on a freighter. He raced stock cars in his native British Columbia before going off to earn a Ph.D. in business at Stanford in the late '60s. McKinnell has known only one company in his post-academic career, and by the look of his resume, he's known only success at Pfizer.

He joined in 1971 and went to Japan. Within a decade he was president of Pfizer Asia. In 1984, Pfizer brought him back to its New York headquarters, where he served at almost every major executive position--president of global pharmaceuticals, chief financial officer, chief operating officer, executive vice president--before rising to CEO in 2001.

To his critics the problem is simple: McKinnell is a doctor of economics, not medicine. As such, he has molded Pfizer in his image, transforming the company from a scientific, research-focused organization into what is essentially a financial holding company with a portfolio of drugs acquired through big deals.

McKinnell contends that those very deals have helped avert catastrophe. In 2000, as president and chief operating officer, he aggressively outbid American Home Products (now Wyeth) to purchase Warner-Lambert for $84 billion, thereby acquiring Lipitor, which has been the world's bestselling drug for five years running. (Last year it accounted for fully a quarter of Pfizer's sales.)

In 2003, McKinnell masterminded Pfizer's $56 billion purchase of Pharmacia. This time the company acquired the full rights to blockbuster pain relievers Celebrex and Bextra. Because of some $50 billion in Lipitor sales, among other things, the Warner-Lambert deal is largely viewed as a success. The Pharmacia deal? It gets low marks partly because of the COX-2 woes.

If only Pfizer's 14,000 scientists were as fruitful as its acquisition team. They haven't been. Investors still consider Pfizer and its peers to be growth companies, but in order to keep growing from its huge base, Pfizer has been forced to compensate for its unproductive labs by wheeling and dealing to buy or license potential blockbuster medicines from smaller, more industrious companies.

McKinnell says Pfizer will continue to acquire other companies on a smaller scale, as the company looks to players with pricetags ranging from $1 billion to $4 billion. McKinnell's open-wallet strategy is fine as a temporary solution to fill gaps in the pipeline, but it is ultimately a more expensive way to "discover" drugs.

Under McKinnell, Pfizer has never achieved a balance between acquiring compounds, the raw material of drug development, and internal discoveries. Pfizer's size may have something to do with its discovery troubles. "Research productivity really doesn't scale with size," says Derek Lowe, a scientist and author of "In the Pipeline," a weblog. "Certainly, if you have eight scientists in the lab and you move that up to 16, you're going to boost productivity. But if you've got 8,000 researchers, discovery won't be any easier with 16,000."

John LaMattina, head of Pfizer's $8-billion-a-year research operation, insists that size matters. "The view that we're big, slow, and dumb isn't quite right," he says. "I'm blessed with having a research organization with scientists in Japan, England, Connecticut, and California. When I grew up in Pfizer, we were small. We had to make difficult choices in research. Do we go after this compound or that one? We didn't have the resources to do it all," says LaMattina. "Now, having enormous size and scale, we can cast a wide net." Indeed, Pfizer's pipeline encompasses almost every major therapeutic area. LaMattina seems only mildly concerned about the discovery drought. Research goes in boom and bust cycles, he says. "I think we're at the bottom of this cyclical curve."

Meanwhile Pfizer hopes to launch six new medicines this year. The biggest potential sellers among them include Exubera, an inhaled insulin for diabetics, and Sutent, a cancer treatment. Each could achieve peak annual sales of $1 billion or more a year, say analysts.

But the brisk launch schedule masks the fact that only one of Pfizer's new crop--Champix, a stop-smoking drug--was discovered within the company's research operation. Giving McKinnell the benefit of the doubt, is it possible that Pfizer has found a new way? Perhaps it doesn't matter where bestselling drugs are discovered. If so, add that to the list of things Hank McKinnell has to prove.

There's a more immediate question: Even with acquisitions, does Pfizer have enough in its arsenal to cover the coming revenue losses? For the short term, yes. But it won't be easy to pull off this year's six launches. "If they stumble with one or two, that will make it very difficult," says Bert Hazlett, an analyst at Suntrust Robinson Humphrey. McKinnell has said Pfizer will resume growth in 2007.

Much of Pfizer's future rests on an upcoming drug, torcetrapib, which is designed to raise good cholesterol, or HDL. The new medicine, acquired by way of Pfizer's $1.3 billion purchase of Esperion Therapeutics, is expected to be combined with Lipitor to form a cholesterol-fighting juggernaut. The new drug--either sold as a combination pill or a separate medication--could generate as much as $3.5 billion a year. It appears to work, recent tests show, and Pfizer expects to file for FDA approval in 2007. But the FDA is under intense public scrutiny and remains extremely risk averse. One yellow flag: Patients taking a formulation of torcetrapib and Lipitor experienced a rise in blood pressure. Says Deutsche Bank analyst Barbara Ryan: "Torcetrapib is the big question mark.... It's not a slam dunk."

Apart from drugmaking, McKinnell has set in motion another controversial plan. Last year he promoted three top executives to vice chairman--Karen Katen, head of sales and marketing; Jeff Kindler, general counsel; and CFO David Shedlarz--clearly signaling a three-way race to become the 13th CEO in Pfizer's 157-year history.

From the outside, the race appears to be running smoothly. But while companies such as General Electric are accustomed to runoffs, it's a practice that is foreign to Pfizer's culture. McKinnell was the single heir apparent for several years before he became CEO. Insiders claim the succession race is creating factions and distracting the company from its larger goals.

"There are David people and there are Karen people," says an insider referring to two of the candidates. "They're pulling in different directions, trying to make their guy look good." Pfizer declined to let FORTUNE interview the three potential heirs. Asked about the succession plan, McKinnell plays down the tensions, insisting that it's not a three-way race at all. "We have any number of candidates who could be the next CEO," he says.

There's no doubt that McKinnell is writing a new playbook at Pfizer. It will be a long time before it's clear whether he got it right. But McKinnell scoffs at the notion that he's fighting to create a legacy. "I've been here for 35 years, and I was part of the team that moved this company from 14th in the world to No. 1," he says. "That's a legacy in itself."