(FORTUNE Magazine) – If you've ever gotten serious (or worried) enough about your financial future to run your stats through a retirement planning calculator, you know what it is to see the future with crystal clarity. These devices range in sophistication from toylike paper slide rules to full-featured software programs. But none of them suffers from doubt. Take, for example, the calculator bundled with Quicken, the popular cash-management software. Input data including the income, age, and current savings of the median FORTUNE subscriber ($97,700, 48, and $492,400, respectively); stir in a forecast of 3% inflation and 10% returns; and Quicken asserts that you can expect to kick back at age 65 on $92,760 a year.

After the initial satisfaction wears off, however, some skepticism should set in. After all, if your forecast is even slightly off--which would still make you more on target than most economists--you'll miss your retirement goal by thousands of dollars. You don't need to be a Nobel Prize winner to recognize a serious garbage-in/garbage-out problem here.

But if you were a Nobel Prize winner, you might actually try to solve the problem--which brings us to William Sharpe, a Stanford economics professor and 1990 Nobel laureate with a side business in corporate pension consulting. Appalled by what he calls the "pseudoscience" of the advice given to participants in 401(k) plans, Sharpe joined forces with two partners to create Financial Engines, a Web-based program designed to give individuals in 401(k)s the same sophisticated portfolio tools used by managers of billion-dollar pension funds. (Disclosure: Among the firm's backers is a venture capital firm in which FORTUNE columnist Stewart Alsop is a partner.) Sharpe & Co. plans to start selling the service to employers--who then make it available to their employees via the Web--around midyear. In the meantime you can test-drive a demo at www.financialengines.com, which is scheduled to open the day this issue of FORTUNE hits the newsstands.

One key innovation: The program builds some uncertainty into its forecasts (hey, just like real life). You input data such as your intended retirement age, the percentage of income you're saving, and the funds you own. The software then runs millions of statistical simulations of your portfolio's returns and calculates the range of sums it is likely to deliver by your retirement date, plus the probability of your retiring with what you need. Don't like the results? Just increase your savings and lower your risk level. As you'd expect, that narrows the range of potential returns and raises the probability that you'll hit your target. Not as simple as promising $92,760, but it's a lot more honest.