(MONEY Magazine) – DOES FRANK SINATRA OWN SHARES IN MUTUAL FUNDS? I confess I've been wondering, since every time I've heard him croon "It was a very good year" lately, I can't help but think he's describing 1995's fund returns. To Dec. 1, the average U.S. stock fund was up 29.9%. The typical bond fund? Up 13.5%. Foreign stock funds? Up 5.2%. You name it, and it was up in 1995.

And that happy fact brings me to this point: A year like 1995 emboldens the 10-percenters, those chirpy financial advisers and press pundits who mesmerize with the mantra that, since the days of Calvin Coolidge, stocks have averaged 10% a year. To be precise, 10-percenters say, from 1926 to last Dec. 1 stocks have gained 10.5% annualized.

My advice: Don't bank on every year being like 1995. That's fairly obvious. What isn't is this: Don't buy the 10% hooey. Remember that the 10.5% annual historical return on stocks is not only an average but does not count taxes, inflation or fees and commissions. Taxes and inflation alone slash the historical return on stocks roughly in half, to 4.3% annually (except for IRAs or 401(k)s, where taxes are deferred). And if you buy stocks through mutual funds, subtract another 1.5%, the average yearly fees that funds charge to run your money. Now 10.5% looks more like 2.8%. Not bad in real terms, but not 10%.

Moreover, stocks return 10% a year only if you bought them in 1926 and sold in 1995--and you can't do that unless you have Marty McFly's time machine in your garage. In 1926, the typical stock sold for 10.9 times its earnings per share and 20 times its dividends. Today, the typical stock sells at 18 times earnings and 44 times dividends; stocks are roughly twice as expensive in 1996 as they were in 1926. How could they possibly return as much for the next few decades as they have since 1926? The answer: They probably won't.

I asked Chicago-based Ibbotson Associates, a keeper of historical returns, to figure out the gains that stocks would have produced if they had traded in 1926 at valuations similar to today's. Using data from 1926 through 1994, Ibbotson estimates that annual stock returns would have dropped from 10.3% to 9.7%.

Why quibble over the difference between 10.3% and 9.7%? Simple. It's because over long periods, such gaps become gulfs. At an annual return of 10.3%, $10,000 grows to $115,981 in 25 years. But at a 9.7% return, it becomes only $101,197; that shrimpy 0.6% annual deficit turns into a shortfall of nearly $15,000. The longer the time horizon and the bigger the gap, the larger the shortfall.

But doesn't history still suggest that you should put all your long-term money in stocks? I don't want to be a party pooper, so listen instead to Roger Ibbotson, the Yale finance professor who founded Ibbotson Associates: "Long investment horizons should definitely point you toward stocks. But I would never say that it's a sure thing. You cannot conclude from history that stocks will always outperform fixed-income investments."

Why not, ask the 10-percenters, especially since there has never been a 30-year period since the 1870s during which stocks failed to whip both bonds and cash? Let me put it this way: Thinking that stocks will always whack fixed-income assets because they always have in the past is like concluding that Cal Ripken Jr. will start at shortstop for the Orioles forever because he hasn't missed a game since 1982. Someday the streak will end. In fact, Martin Leibowitz, who oversees $155 billion as chief investment officer of the world's largest pension fund, New York City-based TIAA-CREF, calculates that there is a 19% likelihood that stocks will underperform bonds over even a 30-year period, based on valuations similar to those in today's market.

Don't get me wrong. I'm not saying you shouldn't buy stock funds; of course you should. But it's not 100% certain that stocks will beat bonds over the long run. So while you should definitely not encase the bulk of your retirement funds in the cement overshoes of bonds or cash, keeping 20% to 40% in such assets is a good idea.

Above all, be realistic. While a 10% return on stocks would be mighty nice, you'd be foolish to count on it--no matter how long your investing horizon is. And don't forget how damaging fund fees can be to the returns you actually earn. If inflation and taxes knock down stocks' mythical 10% return to 3% to 5%, a 2% annual expense ratio devours at least 40% of your take. That's robbery; try to find funds with annual costs of 1% or less. Vanguard's no-load index funds, with their rock-bottom expenses, are the place to start shopping. Other low-cost providers include Dodge & Cox, Fidelity, Mutual Series, Neuberger & Berman and T. Rowe Price. If you use a broker, stick to low-free purveyors like the American funds, Davis funds or Franklin/Templeton--and pay the sales load the old-fashioned way: up front. It'll cost you less in the long run.