(FORTUNE Magazine) – After talking nearly nonstop about his business for more than two hours, Henry Kravis walks a visitor through Kohlberg Kravis Roberts & Co.'s elegant 42nd-floor Manhattan offices to a room called the library. To the right is a stunning view of Central Park. Directly ahead, on the west wall, is what we've come to see: a large, framed graphic filled with logos of KKR deals-Safeway, Duracell, RJR Nabisco, Beatrice-and in the center a yellow deposit slip with which the legendary buyout firm, days after its founding on May 1, 1976, opened a checking account at Chemical Bank with $10,000.

The graphic, a gift from a top investment banker at J.P. Morgan Chase, celebrates KKR's storied past and underscores the firm's indelible mark on Wall Street history. While KKR didn't invent the leveraged buyout, in the 1980s it became undisputed king of the field-the only firm that could have pulled off the $25 billion RJR Nabisco deal in 1989, still the largest LBO on record. But RJR turned out to be a watershed, a costly and damaging miscue after which KKR went through a period of embarrassing sloppiness -misjudging markets and managements-that eroded its competitive position. Says Kravis: "Quite frankly, we got sort of arrogant and made some terrible mistakes." One he has in mind is Regal Cinemas, which lost KKR and Hicks Muse a full $1 billion, a record for the buyout industry. (We give you a rare, revealing look at KKR's performance figures further on.)

Those mistakes eventually spurred Kravis and George Roberts, his partner, first cousin, and alter ego, into action. Around the start of this decade, they put on their boots and plowed into improving the ways in which the firm makes and rides herd on its investments. It's still too soon to judge the success of that campaign-it takes time in this industry for results to come in-but certainly KKR has planted its flag in Europe and generally engaged in a ton of transactions. Since the beginning of 2004, KKR has formed one new specialty printing and marketing company, Visant, and acquired control or a major stake in eight other corporations, including Sealy, PanAmSat, and power generator Texas Genco. And it has been vigorously cashing in some of its holdings, including its entire remaining stakes in Owens-Illinois and Walter Industries (once called Jim Walter), which KKR amazingly had been in since 1987. From all the sales made in 2004, KKR returned $7 billion to its investors. As a bonus, Kravis and Roberts also contend, with some justification, that they have transformed KKR into a best-practices leader.

Even some rivals make a bow to the firm's resurgence. Said one recently: "KKR made so many bad deals for a while that it began to look as if it couldn't hack it." But, he says, "I have to hand it to them, because they've come back." David Roux of Silver Lake Partners, generally a competitor of KKR's, though right now a partner in the big deal to buy SunGard Data Systems (see preceding story), views KKR's ups and downs as an arresting, long-run saga: "They were great in a precocious way; they taught all the rest of us how to do it. Then they went into relative exile, and now, sort of like Winston Churchill, they've come back. It's a fascinating story."

Yet it has played out in ways that show how much Wall Street has changed since the 1980s. KKR no longer rules over the buyout field, which has too many jousting princes these days for any absolute monarch to emerge. And the firm has had to change the way it does business too. Instead of habitually engineering solo acquisitions, it is increasingly joining with other outfits in less risky club deals. Two cases in point: In a deal it first sought to win on its own, KKR ended up in March negotiating a $6.6 billion buyout of Toys "R" Us with two partners, Bain Capital and Vornado Realty. Next SunGard: No less than seven buyout firms, with Silver Lake in the lead, are to buy the company for $11.4 billion, an amount slated to make SunGard the second-biggest deal after RJR.

If sharing the spotlight bothers Kravis and Roberts, they don't show it. Indeed, though both are 61 and rolling in riches, they have absolutely no interest in retiring-they still live and breathe buyouts. That point was made abundantly clear throughout lengthy and unusual conversations that FORTUNE had this spring with these normally press-shy men. You can't miss another fact, either: Both are intense, hard-charging types who love to compete and aren't ready to concede an inch of advantage to their rivals. Says Roberts of KKR: "I still think we're the best guys out there."

The cousins (who split from their co-founder and mentor, Jerome Kohlberg, in 1987) continue to run the firm much as they always have. Roberts, who dislikes New York, has worked for decades out of California-first in San Francisco, now in Menlo Park. So four or five times a day, maybe, he and Kravis are talking business on the phone. The cousins haven't had an argument, claims Kravis, since they were 7 years old (when they fought over who would get to ride young Henry's new bicycle). But devil's advocate each can be: "Usually if one of us loves an investment idea," says Roberts, "the other will take a more challenging opinion."

In their personal lives, both endured losses. Roberts's wife, Leanne, 57, died two years ago. Kravis's son, Harrison, died in a 1991 car accident when he was 19. Two portraits of Harrison, dark-haired and handsome, hang in his father's office conference room. In a better turn for Kravis, he is now happily married to his third wife, Marie-Josée, an economist and corporate director (of Ford and InterActive Corp-and formerly of Hollinger). "The third time around," says Roberts, "Henry got it right."

Kravis and Roberts share certain outside interests-the contemporary art that decorates their offices, for example, and golf. Kravis was captain of his golf team at Claremont Men's College, and if you give him an opening, he will tell you about a great shot he made while playing with Tiger Woods not long ago. Roberts, spurred to take up the sport by his friend Chuck Schwab, has in the past 15 years obsessively gone from rank beginner to a four handicap. Many evenings, after work, he can be found on the practice tee at Stanford University's course, a few miles from his office. Physically, neither he nor Kravis casts a wide shadow while playing: Both are short and slight. But they are almost comically different in character. Roberts is soft-spoken and reserved, almost reclusive. Kravis is gregarious and hugely social. Roberts recalls that when the two were young bachelors and roommates in New York City, Kravis was out every night--a routine, says Roberts, "that would drive me nuts."

WHAT THE NUMBERS SAY

Their divergent social styles-and locations-haven't prevented them from building an exceptional track record. Being a private operation, KKR never broadcasts the details of its returns. But when the firm raises a new fund, as it is doing right now in Europe, it lays out its record in a private-placement memorandum that becomes its selling tool. FORTUNE secured one of these documents, 69 pages long, from a state pension fund that is an investor-a limited partner-in KKR's deals. It provides an illuminating look at KKR's performance.

From its start in 1976 through September 2004, KKR raised ten funds and sank $21 billion of its investors' money (including about $500 million from KKR partners and other insiders) into 93 companies. Counting debt, it put a total of $130 billion into them. On these companies, it figures its profits as of September at $34.7 billion. Of that, $26 billion was realized profits and $8.7 billion was the firm's estimate of profits there for the taking on 28 companies it then held. (KKR realized some of these profits, and more that had materialized, in the six months after September, when it engaged in a large round of successful selling.) As for KKR's lifetime batting average, the document has the firm making money on 62 companies (including one that went bankrupt, builder Jim Walter); losing on 22 (including three bankrupts, Seaman's Furniture, supermarket chain Bruno's, and Regal Cinemas); and square on nine, most of them recorded that way because they have been owned for only a short time. Says Roberts of the $34.7 billion in profits the firm has produced: "That's probably more than our five or six biggest competitors have done combined."

Here's how those profits translate into the two measures by which the buyout industry lives and dies. First, there's "internal rate of return," or IRR, a complex calculation that takes into account the flows of money in and out of an investment. KKR figures its annual IRR over the years to be 27%. Second, there's the relationship of dollar profits to the money put up. Here, KKR through September had generated profits of $1.65 for every $1 invested (though in the way the buyout business works, only a fraction of the $1 would have been invested at any one time).

On the face of it, the figures look impressive, but a closer inspection shows that the long-term trend in the firm's returns is not good. KKR's first five funds (see table) produced huge returns, averaging, on a dollar-weighted basis, almost 37%. In the next four (not including a $6 billion fund raised early this decade, because it has not yet matured), the weighted average fell to 16%.

That's obviously a prodigious drop. Only part of it, though, resulted from unadulterated poor performance-those "terrible mistakes" Kravis owns up to. The remaining part mainly reflects the disappearance of the easy pickings of the early years, when companies could be both bought cheaply and leveraged out of their skin. For example, KKR pulled off its $4.3 billion buyout of Safeway in 1986 by putting up only $132 million of equity, a mere 3.1% of the purchase price. KKR, which has traditionally and sometimes controversially gone in for long holding periods, then clung to a stake in Safeway for no less than 17 years, eventually realizing $7.4 billion in profits-the firm's biggest dollar triumph by far. Deals like that just don't come around much anymore.

It's even more crucial to note that the profits are gross-they don't reflect KKR's cut, or "carry," and certain fees the firm charges. In fact, by any measure KKR does better for itself than for the investors who put up most of the money for its deals-and this is a principal and greedy trait that buyout funds share with hedge funds. In the early days, when KKR could pretty much dictate its terms, its carry was 20% of the realized profits of each deal and 0% of any losses-as close to a free lunch as you can get. Today, because the limited partners rioted for better terms, the take is 20% of an entire fund's returns, which leaves KKR sharing in any losses-a significant change. So what part of that $26 billion in realized profits went to KKR? Hard to say, but $5.2 billion would appear to be a minimum and maybe $5.5 billion the max.

The profits that the KKR folk have taken in from the carry are by far the juiciest part of their dollar story, but not all of it. Like hedge funds, buyout firms also get a management fee, which these days may run from 1% to 1.5% of the funds that investors have committed to put up. Going back to the 1980s and early 1990s, KKR also raked in 100% of certain fees it charged its companies (for simply "monitoring" them and for M&A transactions). But today, because its limited partners got irritated about what KKR was siphoning out of deals the partners were financing, the limiteds get maybe 55% to 80% of those fees. So from the viewpoint of a KKR, taking into account its lowered fees and its need now to share losses, there's been a certain long-term deterioration in the economics. But do not weep for these folks. The funds meanwhile rise in size, which increases both the fees they throw off and the potential carry. So think of the KKRs of the world as earning profits that are "gross" by another dictionary definition: "excessively fat."

Kravis and Roberts's share of those profits has naturally declined over the years as they have admitted other partners-14 of them today, all technically "members" of a limited-liability corporation-into the firm (which, by the way, still has a 1% owner named Jerry Kohlberg). Still, one source who knows the workings of the firm estimates that from the carry and other profits, each of the cousins has received at least $2 billion after tax-and, he says, "I could be low." Furthermore, they have probably banked most of that money: While many hedge fund principals have much of their net worth invested in their funds, that is not normally the practice in the buyout business, which is a reason its firms have to keep hitting the road to raise money.

Why do the limited partners pay the freight? Because they expect the buyout funds to deliver "excess returns" over what they might earn in the stock and bond markets. It's worth pointing out that the "limiteds," as they are called, could very likely drum up at least equal excess returns were they willing to play the leverage game themselves by loading up on S&P index futures. But pension funds, being fiduciaries, don't like to speculate that way. Instead, they line up at the doors of buyout funds.

From the beginning, KKR's key investors-bound to the firm with almost religious devotion-have been the investment arms of two states, Washington and Oregon. Both figure their annual returns from KKR over the years to have exceeded 16%, which they think both entirely satisfactory and a reason for them to dig deeply when KKR periodically brings its collection plate around. Their latest act of fealty occurred early in this decade as KKR worked to raise what came to be called its Millennium fund and ran into two problems: first, resistance from investors because of KKR's recent spotty record, and second, the fierce blow of 9/11. In the end Oregon stepped in with an unprecedented investment of $1 billion, and Washington topped that with $1.5 billion. So of the $6 billion fund KKR raised, two investors amazingly account for just over 40% of its dollars. No problem, says Joseph Dear, executive director of Washington's investment board: "KKR has a demonstrated capacity to show good returns."

THE LEAN YEARS

There was a time when the returns went south, and it started with KKR's ill-starred 1989 acquisition of RJR Nabisco. After a slugging match between the company's management and KKR to see who would take this huge company private, KKR "won," so to speak-handing over the $25 billion. Ultimately, in three swatches, the firm put $3.7 billion of equity into RJR and a related deal, Borden. The $3.7 billion was also a remarkable 60% of the money deployed by the vehicle known as the 1987 fund. But that deal ran into big trouble from tobacco litigation and bond market woes, and the firm's investors lost an ugly $958 million. KKR fared better: It was not then sharing in losses, and it reaped almost $200 million in fees from RJR and Borden. Even so, the results were humiliating.

That is still evident a decade later as Kravis, sitting in his paneled conference room, shares his rueful, highly personal recollections of the deal, among them a feeling that "we were so close to solving our problems." Still, he says, there is lesson No. 1: "Never issue reset notes." As this suggests, RJR had notes that reeked: $7 billion of pay-in-kind securities. Issued in 1989, these notes came with the promise that by April 1991, RJR would, if necessary, "reset" their interest rate at a level that would give the notes a fighting chance to sell at par. But by early 1991 the junk-bond market had collapsed, the ratings on RJR's bonds had been downgraded, and its notes were selling at 70. There was no way a reset could lift them to par. Kravis says one visitor to his office declared bankruptcy inevitable. Kravis said no way. But KKR was forced to put more money ($1.7 billion in that swatch) into RJR to save it.

Lesson No. 2, says Kravis, is "Don't rely too heavily on past outcomes." KKR always planned to split RJR's food division, Nabisco, from its lawsuit-embattled tobacco business. But for tax reasons, RJR had to wait five years to do so. Unfortunately, says Kravis, "in about four years and nine months" there began to be signs that tobacco plaintiffs were gaining ascendancy. KKR's lawyers therefore warned Kravis and Roberts that splitting off the food division might be construed as "fraudulent conveyance"-i.e., wrongful disposal of an asset that might otherwise be used to satisfy the plaintiffs' claims. Worse yet, the two men, as controlling principals of KKR, could have been held personally liable. So KKR didn't split off the food division but instead made a faux exit by trading half of its RJR shares for a piece of Borden. And that company, says Kravis, looking quite pained by this time at this dolorous recital, "turned out to be much weaker than we thought."

By the early 1990s, of course, the RJR saga had been made notorious in Barbarians at the Gate, both the book (co-written by John Helyar, now of FORTUNE) and the movie. (Snap quiz, difficult variety: What actors played Kravis and Roberts?) For Kravis, who cares about being liked-Roberts seemingly can take it or leave it-the firm's unending public relations problems were excruciating. He recalls a friend's calling him on behalf of a client wanting to know who handled KKR's PR. Responding that it was Kekst & Co., Kravis asked, "Why does your client want to know? Does he want to make sure he never uses them?" The caller laughingly said that to the contrary, his client was impressed by how well KKR was faring publicly. Said Kravis: "You've got to be kidding. We're getting killed. I chop my head off every day." (Follow-up: Kekst & Co. is still KKR's PR consultant. In the movie, Jonathan Pryce played Kravis and Peter Dvorsky played Roberts.)

One of KKR's buyout competitors, John Canning of Madison Dearborn, says that the blistering experience of RJR "could have ruined a lot of firms," and that KKR's picking itself up and marching on was testimony to its strength. The firm did in fact make another mark for itself in the mid-1990s, by successfully raising a then-record amount of money, $6 billion, for what came to be called its 1996 fund. But the terms of that fund indicated that KKR's status in the buyout world was weakening: A growing cadre of tough, hungry competitors, working to get in the game, had by then made the sharing of losses and fees commonplace and had forsworn-echoes of RJR!--ever putting an outsized proportion of one fund into a single company. So KKR's limited partners and one prospective investor, Calpers, put the arm on the firm to match the competition. The argument got so heated that Roberts told Calpers to take a walk, saying it wouldn't be welcome in the fund (though years later the two made up). Eventually, KKR capitulated. The firm agreed to lower its management fee (from 1.5% to 1.1%), split its monitoring and transaction fees with the investors, and switch to a loss-sharing system. So suddenly KKR wasn't getting the tribute that it had become accustomed to as king.

The 1996 fund has since grown into what buyout firms call the "harvesting" phase, and is therefore ripe to be judged. In the way that counts most, it has done fine. As of September, the fund was showing a net IRR of 13.4%; by March, according to one limited partner, that had climbed to 14.3%. In a period that included the bubble and in which general expectations have been lowered, that's competitively respectable.

The fund has had some home runs, particularly British insurance broker Willis Group. KKR took Willis private in 1998 in a $1.7 billion management buyout. In 2000, KKR made a key move by hiring Joe Plumeri, a former Citigroup executive, as CEO. Willis then went public in 2001, and it recently had a market value of $5.6 billion.

The weakness of the 1996 fund is that it's had a terrible batting average, like Jason Giambi off steroids. Of the 34 companies KKR put money into, 15 became losers. Some of the flops were telecom companies that KKR, though it likes to think itself a value investor, got sucked into during the bubble hysteria. A hoary struggler is Primedia, which owns about 120 special-interest magazines (Motor Trend, for example). The company has had problems, to put it mildly; says Roberts dryly, "Where can I start?" Beginning in 1989, three different KKR funds have pumped $1.2 billion into Primedia (originally called K-III), which makes it the firm's second-biggest investment after RJR. Last year Primedia unloaded New York magazine for $55 million. Recently it sold About.com-a subsidiary whose value had been written down to almost zero-to the New York Times Co. for $410 million. Primedia's stock, which three years ago hit 85 cents, is now around $3.75, and KKR figures itself about $20 million in the hole on the deal.

For Kravis and Roberts, the most galling losses in the 1996 fund seem to be two that mocked their good record in consumer companies. One was the infamous Regal Cinemas buy, which KKR and Hicks Muse carried out in 1997, at a time when the buyout industry was mad for movie theater companies. But everybody then got wildly aggressive about building new screens, to the point of generally ruining the business. At Regal, says Roberts, management not only overbuilt but also "played one owner against another." Final result: bankruptcy and a wipeout of the two firms' $1 billion in equity.

The other disaster was sporting-goods company Spalding, an investment split between KKR's 1993 and 1996 funds. Hashing over the Spalding story, a rival sporting-goods manufacturer recently said, "KKR ruined it." Kravis seems implicitly to agree. "We thought management was okay," says Kravis, "and we took our eyes off the company." KKR eventually lost more than 90% of its $674 million investment.

RENOVATION

Sometimes it takes pain to get progress. As the 1990s were ending, Kravis and Roberts shared many long phone calls, and decided their firm must be shaped up. Over the years since, they've made five organizational changes, all aimed at differentiating the firm from what Kravis calls an industry that's become "commoditized." But the result of those changes is to make KKR a little less dependent on the swagger and guts of its two principals.

The shakeup started with the two bosses giving up their practice of making all the investment decisions by themselves. Instead they now are members of an investment committee that includes a handful of their senior partners as well. The committee convenes every Monday, with all concerned obligated to make the meeting a priority.

Next to be organized was a portfolio committee, which gathers quarterly to closely examine the performance, well-being, and possible salability of all of the companies in KKR's portfolio (27 at the moment). Kravis and Roberts and partner Perry Golkin are on the committee, but it also includes a string of senior advisors that KKR has hired for their operating expertise. Among the advisors are some former corporate CEOs, such as Ed Artzt of Procter & Gamble and George Fisher of Motorola. A relatively new advisor is retired four-star general John "Jack" Keane, who as vice chief of staff ran the Army's operations. Sizing up KKR recently, he marveled at the "depth of knowledge" that the firm's people seem to have accumulated about every business that KKR is in.

One probable reason for that is another change at KKR: the establishment of 11 industry groups-for example, chemicals and health care. A partner, a managing director, an associate, and an analyst staff each group, and one of their aims is to keep thinking about out-of-favor industries that might yield bargain-priced opportunities. Kravis has told KKR's folk that they are not to learn their industry from talking to CEOs. Rather they are to walk the aisles of trade shows, and by all means get to purchasing agents, who, says Kravis, "will talk your head off telling you who delivers and who has the best product."

The industry experts give KKR a leg up on another change: 100-day plans, to be put into effect immediately upon the firm's gaining control of a company (and to be followed by other 100-day plans). For example, a plan might include the institution of new metrics. Preparing to take over Canada's Yellow Pages Group in 2002, KKR discovered many of its salespeople were making about two sales calls a day. New standards, requiring more calls a day, went into effect immediately. Yellow Pages became a resounding success, resulting in a quick flip and a record 146% IRR for KKR.

The other change for KKR was its reinstitution of an in-house consulting operation. Many years ago an executive named Steve Burd had run such a thing for KKR. But he was pulled off temporarily to be CEO of Safeway, a job he's so far held more than 12 years, and in-house consulting died. Now it's in the hands of Dean Nelson, 46, formerly at Boston Consulting Group. And guess what? He's been pulled off to help out at Primedia. But he has also been busy at Sealy, which KKR bought last year from Bain Capital (a deal exemplifying today's highly popular practice of one buyout firm selling to another). And 18 partners and associates of the consulting operation, called Capstone, are working around the U.S. and in Europe. As for Capstone's pay, it comes from both KKR (which may, for example, want Capstone to do due diligence on an acquisition it's considering) and the companies that KKR controls.

It is too early to tell how much all of KKR's shaping-up exercises will do for it. But the results of its first European fund, which began to make investments in 2000, looked decent through last September, showing a net IRR of 14.7%. KKR, in fact, believes it has made a wonderfully successful entry into Europe, beginning with deals (such as Willis) made before the firm actually opened an office there, in 1999. That operation, in London, was first run by Edward "Ned" Gilhuly, 45, who moved from KKR's Menlo Park office to get Europe off the ground. Today it is run by Johannes Huth, 45, a German national who used to work for Investcorp.

Gilhuly and two other partners, Michael Michelson, 54, and Scott Stuart, 45, are logical candidates to succeed Kravis and Roberts one day. They were, notably, the three men who joined Kravis and Roberts on the investment committee when it was formed. And Stuart has led KKR's formation of a new offshoot, KKR Financial-the first extension of the brand. There would as well be an odd symmetry to Gilhuly and Stuart's getting the nod, because they were roommates at Stanford Business School, are the same age, and do the two-coast routine, Gilhuly working in California, Stuart in New York.

But "eventually" is the operative word, because there's no sign of a spot opening up at the top anytime soon. In fact, the terms of the European fund now being raised provide that investors can pull out money if either of the cousins stops giving KKR the major part of his time. Not to worry. Roberts guffaws at the very idea of retirement. In March, Kravis got so wound up discussing his business that he stopped to half-apologize to his visitor. "You may be asking yourself why I'm so fired up," he said. "I love change. I love challenges. When we can go in and fix a business, I really love it."

You have to believe him, because why else would this 61-year-old multibillionaire recently have taken a three-week business swing around Asia? It wasn't easy, not even with the red-carpet treatment that Kravis gets. The press of people in China, he said, was very tiring. But he wanted to do three things there: The first was to investigate how KKR (whose companies already have 35 plants in China) might "source" its own operations better. The second was to meet with Chinese companies that might want to invest in the West (by, say, partnering with KKR on deals). The third was to consider Chinese investment opportunities.

And he found, he said with satisfaction, that Chinese businesspeople know the KKR name well. One executive told Kravis that for training he had his people watch Barbarians at the Gate. Whoa, said Kravis, maybe that's not a great idea. Oh, yes, said the executive, "I want them to take risks. I want them to understand the sky's the limit." That was a thought Kravis couldn't exactly argue with, so he didn't.

29 Years, $34.7 Billion in Profits and Counting

Here's a look at the performance of the ten funds KKR has launched.  The first four have shed all their holdings.  The next three are largely out of theirs, and the 1996 fund is busy selling.  The last two are still investing.

The KKR Companies

The roster changes regularly; here's the latest lineup.

FEEDBACK cloomis@fortune.com