NEW YORK (CNN/Money) - Warren Buffett is the rare corporate chief willing to shine a spotlight on his mistakes.

He's had a lot to focus on in recent years. In 1999, it was the inferior performance of Berkshire's stock portfolio. In 2000, soaring marketing spending at auto insurer Geico failed to generate the promised customer growth. In 2001, reinsurer General Re recorded enormous losses, mostly as a result of the Sept. 11 terrorist attacks but also because of more endemic underwriting mistakes.

"I let Berkshire operate with a dangerous level of risk," conceded the Oracle from Omaha.

Such straight talk, coupled with exceptional long-term returns, helps explain why Buffett is so venerated. Amid a punishing market downturn, it's a combination that investors have embraced, with Berkshire's A shares rising some 75 percent since early 2000, compared with a 23 percent loss for Standard & Poor's 500-stock index.

Even Janus Capital, a fund family known more for chasing tech stars than insurance stocks, has been gorging and is now the largest owner of Berkshire's B shares, which represent 1/30 of an A share.

And shares have rallied recently, as the company has posted encouraging results suggesting that business has stabilized (see "Insurance gains boost Berkshire"). But the decision to buy Berkshire stock isn't the no-brainer it used to be. MONEY magazine examined some of Berkshire's largest businesses -- General Re and Geico -- as well as the $28.7 billion equity portfolio and found that, for all the stock's strengths, there are greater risks than you might think.

In the past five years, Berkshire has changed from being primarily an investment portfolio to an emerging conglomerate (a la General Electric), with insurance activities accounting for 55 percent of its revenue. The cornerstone of this transformation was the 1998 purchase of General Reinsurance, a kind of insurer for insurers -- it takes on the risks of catastrophes that most insurers lack the capital base or the inclination to handle.

The reason Buffett loves the insurance business is that its premiums provide buckets of cash, known as the float, for him to invest. The strategy, as with all insurers, is to keep claims and underwriting expenses -- the cost of that float -- low, while earning hefty returns on it.

In General Re, Buffett bought himself $15 billion in new float, tripling the size of Berkshire's war chest. But soon after the acquisition, unpleasant surprises surfaced. Seems General Re hadn't been charging high enough premiums to cover claims, leading to a $1.2 billion loss in 1999 -- its worst underwriting result in 15 years -- and another $1.2 billion loss in 2000.

After the terrorist attacks of Sept. 11, it became clear that General Re, and the industry, had written policies that covered scenarios insurers had simply never imagined: Berkshire's reinsurance companies suffered a pretax underwriting loss of $2.4 billion, $1.9 billion of it at General Re.

Fast-rising insurance rates should help recoup some of that loss, but don't expect the strong pricing cycle to last. Buffett himself said last November that the pricing power will "almost certainly end within a year."

In any event, this is not the kind of performance investors have come to expect from a Buffett-led company -- particularly in insurance, his core area of competence. He has always embraced the field as more understandable and predictable than, say, tech, an arena he has steadfastly avoided.

General Re isn't Buffett's only misstep. Geico is the nation's sixth-largest auto insurer, selling policies via the Internet, phone and mail rather than relying on agents. It enjoyed rapid growth -- until market leader State Farm launched a price war to regain auto market share. Suddenly Geico had trouble winning new customers. Despite nearly doubling marketing expenses from 1998 to 2000, customer growth stumbled.

"Geico tried to grow too fast," observed a rival, Allstate CEO Edward Liddy. In fact, last year it cost Geico $213 on average to lure a new customer, up from $147 in 1999, while the number of policies on the books fell 0.8 percent.

The battle has abated recently. State Farm, after a huge 2001 loss, is aggressively raising rates, which smoothed the way for Geico to hike prices without losing market share. Still, low-cost provider Geico is under assault as more and more rivals move into direct sales. It needs a new strategy to revive its growth -- but we haven't seen it yet.

Most folks, of course, don't hold Berkshire because they want to own an insurance company; they want to tap into the master's investing acumen. Buffett's gift for buying consumer stalwarts on the cheap is the reason Berkshire boasts a massive portfolio of stocks such as American Express, Coca-Cola and H&R Block. But as with the insurance business, recent times haven't been kind to Buffett -- despite the performance of his own stock.

Through April, his stake in Coke has sunk 18 percent since 1998 and Gillette is down 27 percent; his American Express stock is down 25 percent since early 2001.

Things seem unlikely to get better soon, as Buffett himself admits. He told shareholders in February that Berkshire's portfolio isn't undervalued. With strugglers Coke and Gillette accounting for 45 percent of the holdings, that may be typical Buffett understatement. Indeed, he's said that he is "decidedly lukewarm about the prospects for stocks in general over the next decade."

Bad news on two counts: Berkshire's stock collection won't see the kind of gains racked up during the recent long bull market. And Buffett & Co. will have a harder time wringing returns out of its insurance premiums. (In 1999 and 2000, investment gains were the only thing keeping Berkshire's insurance division out of the red.)

"His expectation for much lower equity returns throws a big part of the reason for investing in Berkshire right out the window," says Hilliard Lyons analyst John Roberts. "In essence, he's saying, 'Don't buy my stock.'"

Of course, there are other places to put Berkshire's money besides stocks. Last year saw the acquisitions of carpet maker Shaw Industries and building-products firm Johns Manville. Yet Berkshire's increasing girth makes it tougher to reach its stated goal of 15 percent annual growth with deals like these.

Writes Buffett: "Some years back, a good $10 million idea could do wonders for us ... Today the combination of 10 such ideas and a triple in the value of each would increase the net worth of Berkshire by only one-quarter of 1 percent. We need 'elephants' to make significant gains now -- and they are hard to find."

Let's give Buffett the benefit of the doubt: Assume he rights Gen Re and Geico and defies the odds to find enough elephants to keep his investment herd moving. There's still another risk here, and it has to do with Buffett's age. While the 71-year-old has joked he'll retire from Berkshire "five or 10 years after I die," clearly his age is no joking matter for investors.

Buffett has a succession plan (his job would be split by two people: one running Berkshire's investments, the other overseeing the operating businesses), and there are internal candidates, but the man is so closely associated with Berkshire that word of his death -- even rumors of illness -- could cause the stock to plummet. How much? Neuberger Berman fund manager Basu Mullick puts the "Buffett premium" as high as $10,000 per share.

It's not just the loss of Buffett that's worrisome: Several of Berkshire's operating chiefs are in their seventies. And Charles Munger, Berkshire's vice chairman who runs subsidiary Wesco, is 78. Even Lou Simpson, slated to take over Berkshire's investments after Buffett's gone, is 65 -- traditional retirement age.

Alice Schroeder, the Morgan Stanley analyst who follows the company most closely on Wall Street (and owns the stock), predicts that a growing economy will unleash the tremendous earnings power in Berkshire's operating businesses and that insurance pricing will stay strong.

She figures that Berkshire Hathaway, recently at $77,500 is significantly undervalued -- worth from $89,000 to $104,000 a share -- assuming that the amount of insurance premiums available to invest nearly doubles over the next decade and that Berkshire earns from 5.2 percent to 13 percent on the money annually.

But Legg Mason Value Trust manager Bill Miller doesn't believe that Berkshire will outperform the S&P 500 during the next several years. (It's now trading at more than 34 times estimated 2002 earnings per share, or a good 50 percent premium to the S&P's average price/earnings ratio.) Miller sold the last of the fund's shares in the first quarter; his No. 2, Nancy Dennin, explains that the stock "no longer trades at a discount to its long-term intrinsic value."

So what's an investor to do? Neuberger Berman's Mullick, who owns some Berkshire shares, tempers his enthusiasm with some legitimate concerns:

"Clearly, Warren Buffett is getting older, and the valuation of the stock is not bargain basement." He can see a 20 percent upside for the stock, perhaps, but those risks also keep him from making Berkshire a dominant holding.

That may be the right kind of balance. Buffett's awesome record has been built in part on his practice of buying stocks at beaten-down prices. He has also proved that holding on through the broader market's consistent over- and underreactions also makes sense. Perhaps the best question to be asking yourself as you ponder Berkshire's stock is: What would Warren do?