Some advisers believe the concepts of withdrawal rates based on historical models make for great talking points, but ultimately should be taken with a grain of salt. "It's like a military strategy game where you're winning wars that were fought and not those that will be fought," says David Diesslin, a principal with the financial planning firm Diesslin & Associates in Fort Worth, Texas.

For Diesslin, handling retirement account distributions is as much an art as it is a science. The art comes from knowing your clients and understanding their fears and needs. Retirement can be scary for many people, because their success is no longer based on their ability to earn but rather on the ability of their assets to earn. "Most people want a run rate they can count on and not have to go back from under normal circumstances," Diesslin says. "But it's not that big a deal for people who are prepared for it."

And that's where the science comes in. Like many advisors, Diesslin uses Monte Carlo analysis tools to build a model that accounts for numerous potential variables that could affect a client's portfolio in the future. He believes these metrics help clients better understand the limitations of their portfolios, and that in turn helps them deal more confidently with the unknown. It also helps them adjust their spending to the realities of current market conditions.

But Monte Carlo is useful up to a point, says Michael Dubis, president of Touchstone Financial in Madison, Wisc. "It gives you thousands of variations, but there are infinite variations that can happen," he explains. "How can you cover infinite variables?"

Because possible future events are boundless, Dubis sticks to withdrawal rates of 3% to 5%, with a preference for rates that are 4% or under. "I don't think I'm being conservative," he says. "I think I'm protecting against the opportunity cost of being wrong."

According to both Bengen and Guyton, withdrawal rates aren't necessarily market neutral because they can be manipulated slightly to accommodate varying conditions. But extreme markets can skew perceptions about what's an appropriate rate, as Ray LeVitre discovered during the late-1990s boom years.

LeVitre, a certified financial planner with Net Worth Advisory Group in Salt Lake City, had some clients who retired early during those heady days and took their 72(t)  IRA early distributions at one of the two higher rates allowed. "I got caught up in the market in '98 and '99 and I didn't put my foot down enough when they spent too much," he says.

He set withdrawal rates of 7% to 9% in a couple of cases. After the market crashed, the IRS gave people the option to make a one-time change to their 72(t) rates, and LeVitre adjusted those clients down to a lower distribution rate. "I think we all learned from the experience," says LeVitre, who adds that he partly relied on the faulty math used by Peter Lynch.

These days, LeVitre sets withdrawal rates by plugging clients into a financial planning calculator on software he owns, and he uses Monte Carlo to gauge various probabilities based on various withdrawal rates.  In most cases he and his clients are comfortable with likely success rates of 85% or more that the money will last for 30 years, and the initial withdrawal rate usually falls between 4% and 5%.

Like many planners, Armond Dinverno of Balasa Dinverno & Foltz in Illinois approaches withdrawal rates by asking clients a simple question: What do you want your retirement to look like?