(FORTUNE Magazine) – Washington Mutual not long ago had all the makings of an epic growth story. Starting in 1996, it transformed itself from an obscure Seattle thrift into not only America's biggest mortgage bank but also its seventh-largest financial institution, boasting $280 billion in assets. In a half-dozen years it purchased no fewer than ten major lenders coast to coast, from Great Western in California to Dime Bancorp in New York, helping fuel an industrywide merger trend. WaMu also mounted a massive drive into retail banking, creating 2,400 customer-friendly branches featuring videogames and play areas for kids and staffed by wisecracking, khaki-clad clerks who called cash "wamoola."

WaMu's bosses radiated hubris. CEO Kerry Killinger predicted that the bank would become the "nation's leading retailer of financial services" and vowed to grab an incredible 20% of market share in mortgages. The chest thumping peaked a year ago when WaMu ran a startling half-page ad in the Wall Street Journal and other papers touting its stock. The ad showed a Roy Lichtenstein--style cartoon of a woman putting on lipstick, with the message: While lipstick colors are mere fads, owning WaMu stock will "look good on anyone."

But WaMu hasn't fully delivered on its promise to diversify: More than 60% of its business still derives from mortgages. And with the mortgage boom waning, it is becoming evident that the company lacks the skills to match its galloping ambition. On June 28, WaMu issued an extraordinary press release announcing a huge downward adjustment in its earnings forecast. It blamed the bad news on "expectations for a sustained increase in long-term interest rates." Huh? How could one of America's biggest mortgage lenders claim the upward march that everyone knew was coming, that Alan Greenspan had been heralding in neon red, was somehow a surprise?

For the banking world the real surprise is the extent of WaMu's woes. Rising rates have exposed WaMu's shortcomings in the bankerly disciplines it needs to handle the wrenching swings of the mortgage business. Critics like Angelo Mozilo, CEO of rival Countrywide, sees it as a classic case of an acquisition-mad business that forgot about the basics. "WaMu was touted as a world-beater," he says. "But it grew too fast. It didn't have the organizational or intellectual capacity to run a business of that size."

WaMu suffers from two maddening problems of its own making. First, it didn't push hard enough to integrate its many acquisitions. As a result, it is saddled with a hodgepodge of outdated computer systems that cause delays and confusion in closing loans. The lack of integration has also prevented WaMu from reaping the savings it needs to make those acquisitions--which cost a total of $26 billion--pay off for shareholders.

It has failed to master a second crucial part of its business as well: hedging interest-rate risk. By skillful hedging, lenders like Countrywide, Wells Fargo, and Fannie Mae are able to smooth earnings through interest-rate cycles, producing strong profits even in periods of tight money like the one we seem to be entering. But WaMu's hedging, at least so far, hasn't worked that way. "WaMu's hedges have been paying off when rates fall, so that they generate big, excess profits," says David Hendler, an analyst with CreditSights, a New York City research firm. "When rates rose, they got killed."

Those problems are hammering profits. The June 28 release predicted that earnings would fall from $3.9 billion in 2003 to around $2.5 billion this year, excluding $400 million that WaMu collected from selling its consumer-finance unit. The mortgage-banking business is the main culprit: WaMu predicted that unit would swing from $1.3 billion in profits last year to virtually zero for 2004.

Take a closer look at the mortgage business, and it's easy to see how profits have evaporated. Start with the computer systems. In 2000 the company announced that it was creating a highly advanced, proprietary platform to handle most of its home-loan operations. It developed the system with Accenture and named it Optis, meaning "best." By late 2002, WaMu's techies were so confident that they dismissed Accenture. The project was subsequently beset with bugs and delays. Optis worked well for processing mortgage applications, but it failed in the far more important tasks of processing, underwriting, and closing the loans. According to sources familiar with WaMu's IT operations, the company started shutting down Optis in March; the snafu was a big contributor to WaMu's $248 million of charges in recent quarters. The company has never acknowledged that it is eliminating most of Optis, on which it is said to have spent more than $500 million, or divulged the amount it will eventually have to write down.

Today WaMu still uses an amazing nine separate systems to originate loans, including one called LoanWorks that it has employed since its days as a small thrift and others that it picked up from acquisitions. The systems have great trouble talking to one another. According to former executives, the Tower of Babel situation drove up costs and led to long delays in closing mortgages this year and last, allowing competitors to poach loads of business. For example, early this year WaMu was taking so long to process loans that it frequently had to redo appraisals, which are good for only 90 days.

The systems confusion caused a blowup last year. Like all banks, WaMu lets borrowers lock in their rate until the mortgage closes. It's critical to hedge that exposure, because rates can rise before the loan is funded, forcing the bank to take a loss. But last year, as rates began to tick up, WaMu's antiquated systems lost track of the locks the bank had granted on loans in its pipeline. In the third quarter, with loan originations still booming, WaMu booked a $126 million loss on its mortgage business because of its setbacks in hedging. "Rule 101 in mortgage banking is managing the risk in your pipeline," says Anthony Hsieh, CEO of the HomeLoanCenter.com, a California lender. "It was very surprising to see a big player report a huge loss on something so basic."

CEO Killinger says that WaMu delayed integrating its acquisitions during the boom so that it could focus on writing more loans--a strategy that generated huge profits until last fall's meltdown. "This integration has simply taken too long," he says. "We didn't execute our normal game plan." He says WaMu is working to unify its systems and plans to be down to five platforms by year-end.

Software upgrades won't do anything to fix WaMu's other major problem: the faulty hedging of its gigantic servicing business. The bank services some $770 billion in home loans for five million Americans, mailing out monthly statements, cashing checks, and escrowing taxes (the writer is a satisfied customer, with two WaMu mortgages). It collects about $400 a year per mortgage for the work.

Humdrum as it sounds, loan servicing can be highly lucrative, especially when rates rise. Rising rates prompt customers to hang on to their mortgages far longer than the five to seven years that bankers typically assume--generating extra years of servicing income for the bank. But servicing is also highly volatile: If interest rates fall, customers prepay their old mortgages, and loans that the bank thought would keep producing income disappear years sooner than expected. Banks protect against such risk by hedging; the practice dampens big swings, preventing losses when interest rates fall at the expense of trimming profits when they rise.

That's the way it's supposed to work. But when WaMu announced its second-quarter results this year, it revealed that its hedging strategy had gone horribly wrong. As expected, rising rates had indeed caused a big increase in the value of the servicing business: $1.7 billion. That should have been enough to cover WaMu's hedging costs and still contribute money to the bank's bottom line. But amazingly, hedging had generated a $2.4 billion loss. It totally wiped out the servicing gain, as well as the bank's income from originations, leaving WaMu with a $63 million loss from mortgage banking.

So glaring was the problem that analysts scratched their heads. Some wondered whether WaMu had been using hedges not just to ensure its servicing portfolio but to speculate on interest rates. Says Jonathan Gray of Sanford C. Bernstein: "It's as if they bet that the Fed would ease and rates would fall. The Fed did everything but take out a billboard saying that it would tighten. Isn't that bizarre?" WaMu has always maintained that it doesn't bet on interest rates. "I apologize for the performance of the mortgage-banking unit," Killinger told analysts. "We will fix it." He added that WaMu has hired BlackRock, the New York City investment firm, to do a thorough review of its hedging practices. In July, WaMu also named a new head of mortgage banking and said it was searching for a new mortgage-banking CFO.

Remarkably, WaMu's fumbling hasn't blasted its stock price, which at around $39 a share is not too far from November's all-time high of just under $47. The shares are holding up, analysts say, because WaMu is an excellent takeover candidate (the 4.5% dividend yield also helps). The lure is its network of retail branches, concentrated in high-growth states like Texas, Florida, Georgia, and Nevada, and WaMu's flair for attracting retail customers. In the rush to build megabanks with national networks, those branches could prove highly attractive to a big bank with an undersized branch system. Both Citigroup and HSBC are rumored to be interested. Killinger says WaMu has no desire to sell, but unless the bank fixes its problems fast, it could become part of the merger trend it helped glamorize.

FEEDBACK stully@fortunemail.com