(MONEY Magazine) – You hear professional investors (including several MONEY 100 fund managers) say it all the time: "We try to buy companies selling for less than their intrinsic value." Sounds good. But what is intrinsic value, anyway?

The phrase was coined by Benjamin Graham, a Columbia University business professor who, with David Dodd, wrote the oft-cited but seldom-read 1934 investing bible Security Analysis. "In general terms," they wrote, "it is understood to be that value which is determined by the facts, e.g., the assets, earnings, dividends, definite prospects, as distinct, let us say, from market quotations." In other words, intrinsic value is the value of a company's business, not its stock. And that figure was worth knowing, said Graham, because while stock prices are unpredictable over the short term, in the long run they will be anchored to a fair valuation of the assets they represent.

Graham conceded that it was almost impossible to calculate intrinsic value precisely. But precision wasn't required in a market bursting with bargains. For instance, in a 1934 article for Forbes, Graham cited White Motors, which had a book value of about $36 million, including $8.5 million in cash. In the profoundly pessimistic market of the time, White's 700,000 shares were going for $7 and change, giving the company a market value of around $5 million--$3.5 million less than the cash it had on hand. You didn't need to know White's exact intrinsic value to see that its stock was cheap.

The devil is in the details. These days, of course, undervalued stocks are not nearly as obvious. So value mavens try to figure the intrinsic value of companies with as much precision as they can. How do they do it? Ten different value investors will likely have 10 methods. Most managers, in fact, employ more than one. Third Avenue Value manager Martin Whitman uses a complex approach that tries to gauge a company's ability to generate wealth not only through its earnings, but also through other means like selling stock or refinancing debt. Dick Weiss of Strong Opportunity likes to compare companies he is looking at with firms in the same industry that have recently been bought out. Others look at liquidation value--what a company's component parts might bring if they were sold off.

A more technical approach is to estimate "discounted cash flow." Robert Hagstrom, manager of the Legg Mason Focus Trust and author of The Warren Buffet Way, says this method (used by Buffet himself) consists mainly of pricing a stock the same way one would price a bond. Bonds pay a fixed return over time. Likewise, a share of stock can be considered ownership of a stream of cash. Calculating the value of that stream means projecting how much money a business could generate over a given period of time and then translating or "discounting" that amount into current dollars.

Garbage in, garbage out. Ultimately, all of these methods rely on assumptions that can turn out to be far off the mark. "You have to make all sorts of projections and estimates," says Weiss, "so it's very much garbage in, garbage out."

Gerry Bollman, a C.F.A. whose Intrinsic Value Associates sells research to clients like GE Investments, estimates intrinsic value by plugging a company's financials into an elaborate computer model. He calculates 25 "intrinsic value drivers," including sales growth, profit margins and cost of capital. He then estimates how much cash they will generate given a company's likely growth rate, which he usually simply extrapolates from the recent past, although sometimes with highly subjective tweaking.

One company he follows is AlliedSignal, which he admires but considers slightly overvalued at its late April price of about $60. Over the past few years, the chemical and aerospace giant has seen its sales grow by about 12% one year and then shrink by close to 3% the next. Bollman's analysis assumes an average annual sales growth rate for the company over the next several years of 6%. On that basis, he estimates that its per-share intrinsic value is in the low $50s. Could he be wrong? Sure, he admits. And a difference of a single percentage point in forecasting that one factor can affect his final intrinsic value estimate by about 20%. "Obviously, it's very difficult to look out even two or three years with some companies and even have a clue," he says. "But if you don't go through the exercise of actually trying to make that forecast, you haven't done your job."

But does it work? Even when you do your job, does value investing pay off? The gospel holds that a portfolio of sound, cheaply bought stocks will, over time, outperform a portfolio selected by any other method. Several academic studies over the years have disputed that premise, though, and the recent returns of most value funds have not been encouraging. Still, the idea that stock prices must be tied in some way to underlying fundamental value remains intellectually compelling. Patience and not a small measure of faith are required. Value investors like to take courage in the aphorism from the poet Horace that is cited by Ben Graham in the first edition of Security Analysis: "Many shall be restored that are now fallen and many shall fall that now are in honor."