I never thought much about retirement until I started opening my monthly individual retirement account statements and saw numbers that made my eyes pop. A few more years of this, I thought, and I could be on a beach somewhere, writing the great American novel. How big would those accounts really have to get before I could seriously consider retiring?

Answer: A lot bigger. Numbers that look good on an IRA statement don't necessarily impress the cold-eyed calculation engines in the three software packages and the three retirement planning Web sites described here.

Anyone musing about the idea of retirement would do well to consult any of these programs, all of which perform the complex calculations--Am I saving enough? How long will my savings last? How long until I can retire?--that will let you know whether you're on track.

With one exception--the ESPlanner software--these tools are variations on the same theme. You tell the program how much money you've got in IRAs, stocks, a pension and so on, how much you make now, when you'd like to retire, how long you're likely to live and how much money you think you'll need to live on. The program crunches the numbers and tells you whether you're in good shape or not. If not, it advises you how much more you ought to be saving.

The biggest difference between the programs is sophistication--software generally allows for more power and complexity than Web site calculators, plus they save your plan and let you come back to tweak it over and over; most Web sites don't do that.

Remember the old computing maxim: garbage in, garbage out. These calculators are only as good as the data you feed them, and it's a good idea to be conservative.

I invented a $500,000 stock portfolio and fed it into a terrific little calculator at SmartMoney.com. Pretend I plan to retire tomorrow and expect to live another 50 years. At 3 percent inflation, I could pull out almost $22,000 a year in inflation-adjusted dollars for the rest of my life. But at 6 percent, that would drop to only about $12,500. Ouch.

The moral: If a program tells you you can retire tomorrow, fiddle with "what if" scenarios to see what would happen if conditions turn out to be worse than you expect.

SOFTWARE

Sometimes computer software can be a wonderfully useful tool. Sometimes it can be the cyber equivalent of a willful two-year-old. Example:

Me: Put this IRA in my retirement plan.

Quicken 2000: No.

Me: OK, please put this IRA in my retirement plan.

Quicken 2000: No!

Me: Put this $#@&! IRA in my retirement plan!

Quicken 2000: No! No! No!

Like any competent adult, I eventually worked out a truce, and in the process learned something important about this latest version of Quicken's feature-rich and generally quite useful money-management program. It's not a retirement planner, at least not in the sense the other programs described here is.

Instead, it is a collection of tools that let you do a lot of things -- keep(BALANCE?) your checkbook, manage your investments, plan your retirement -- as long as you do them Quicken's way. For example, in this case, that means you exit the retirement module, go to the investment module and enter your IRA there. Then you go back to the retirement module and there it is. Too bad they don't tell you that in the first place, but now you know.

Once you get the hang of it, Quicken's retirement planner works well. Its best feature is a powerful "what if" analyzer that makes the process easier than competing programs. Once you've entered all your numbers and found out whether your retirement plan works or not, you can easily change any underlying number and instantly see the effect. This can be very sobering. On my first pass, the numbers allowed me to retire comfortably at 62. But if I assumed 4 percent inflation instead of 3 percent, my plan failed. Ditto if the rate of return on my investments was 1 percentage point lower than I forecast.

On the negative side, Quicken offers almost no guidance on the crucial matter of what rate of return you can expect on your investments, a number that can wildly skew your retirement plan. To wrestle any advice from Quicken, you have to burrow down through the "assets allocation" section to the "model portfolios" page, where you can see that various mixes of stocks, bonds and cash generally yield 6.5 percent to 11.5 percent over time. Which should you pick? You're on your own.

Worst, though, is a problem that Quicken shares with virtually all other retirement planners:It demands to know how much you'll be spending once you quit working. Methods for picking a number range from taking a percentage (generally 60 percent to 100 percent) of what you're spending now to laboriously entering, item by item, how much you think you'll be spending at a certain point in the future.An interesting alternative approach comes from a new program called ESPlanner, (download at http://esplanner.com) developed by Boston University economist Laurence J. Kotlikoff and two colleagues. It focuses on using your savings and income to show you how much you'll be able to spend in retirement.

The results are a bit overwhelming.. ESPlanner generates 28 separate reports, beginning with bare-bones recommendations and quickly escalating to detailed spreadsheets that show how much savings you'll have, what your taxes will be, how much you can spend and so on and on, year by year for as long as you expect to live. There are explanations for all these numbers, but they are written in a sometimes numbing combination of English and econo-ese.

"We're trying to make it as user friendly as possible, but we're economists," Kotlikoff said. This is a feast for the detail-oriented, and if you demand the deepest and most powerful planning engine, look no further. But this might be too much for the average would-be retiree.

A cheaper but quite viable alternative to its two pricier competitors (both Quicken 2000 and ESPlanner sell for $49.95) is the $9.95 T. Rowe Price Retirement Planning Analyzer, (download at www.troweprice.com) a straightforward retirement planner that offers reasonable advice on inflation and investment returns and walks you through a painless interview to produce a readable and useful retirement analysis.

I gave this program extra credit for recommending that I use a conservative 4 percent inflation projection. Unlike Quicken 2000, it said my retirement plan worked at 4 percent, which was gratifying but confusing, since one of them has to be wrong. All the more reason to treat these calculations with caution.

WEB SITES

Leave it to the Web to boil retirement down to the fundamental question: "What's for dinner on your 80th birthday?" asks the Optimal Retirement Planner Web site. On one side, a depiction of a sumptuous meal with shrimp, roast turkey, pie and a bottle of wine. On the other: a can of dog food.

Social Security was supposed to prevent us from having to spend retirement subsisiting on Alpo. But the demographic pressures building on the federal retirement plan as the baby boom approaches retirement age, and indeed the growing retirement anxieties of the aging boomers themselves, have spawned a zillion Web sites devoted to helping you figure out how to plan ahead.

No one will mistake these Web-based calculators for their more complex software brethren, but they are extremely useful, and on many sites they are surrounded with all sorts of helpful articles and advice. Best of all, they are free. Here are three I looked at, ranked in purely subjective order:

www.Quicken.com/retirement

The big software maker also provides the best of these Web sites. I actually liked SmartMoney.com's retirement planner a little better, but Quicken won by saving my data so that I could come back and fiddle with the numbers without having to enter them all over again -- a huge advantage.

That said, this is a pretty good calculator in its own right. In some ways it's even better than its $49.95 software cousin, Quicken 2000. For example, I was frustrated with the software planner for making it hard to find guidance about what rate of return I should use on my investments -- a number that can make an enormous difference in whether your retirement plan is realistic or not.

On the Web version, right at the spot where you enter your expected rate of return, there's a helpful gizmo that shows you the implications of what you're doing. Very nice.On the other hand, while the Web version has a good "what-if" analyzer that allows you to recalculate your plan by changing different numbers, it pales in comparsion to the much more powerful version in the software.

www.smartmoney.com/ac/retirement/

This site has an extremely helpful planner, but it would get big points if it offered nothing more than the elegant little calculator on its opening retirement page. Enter the size of your savings, how many years you'll need to draw on them and bingo: The calculator instantly shows you the maximum amount of money you can pull out each year.

The calculator assumes a 3 percent inflation rate and a 7 percent after-tax rate of return, but you can change either or both of those. I got hooked on this gizmo and used it to run through dozens of different scenarios, testing the effect of lower rates of return and higher inflation (very bad) and much shorter retirement years (good, from a fiscal point of view, at least). For example, if I retired tomorrow with an imaginary $500,000 stock portfolio and planned to live another 50 years in an era of 4 percent inflation, I could draw almost $18,500 a year. If I was only planning to live another 10 years, I could draw nearly $57,000 a year.

By the way, this site, like most other Web-based planners and the software versions, advises would-be retirees to think about not counting on full Social Security benefits, especially if they're more than 10 or 15 years from retirement. While the idea of cutting retiree benefits gets little public support in Congress, the notion of its inevitability is so widely held outside Congress that it has become a regular feature of most retirement calculators.

moneycentral.msn.com/retire/home.asp

Disorganized and not as helpful as its competitors, this site struggles despite some good features and pithy insights from financial planner Ginger Applegarth: "So you thought buying a house was expensive? Think about this -- It's going to cost you four to 10 times as much as what you paid for your home to ensure yourself of a comfortable retirement."

The highlight of this site for me was the life expectancy planner, which is far more thorough than the ones I found elsewhere. On the other hand, it told me I'd only live to 80, and it had the cheek to suggest an ideal weight that was less than I currently weigh. So who asked?

This site's retirement expense calculator was the first one to convince me, step by step, why I'd be spending less in retirement than while I'm working, even if I decided to travel a lot. And its planner was the only one that let me choose my plan from three alternatives that depended on how much I wanted to leave my heirs after I died.