For most retirees, the greatest fear is not death—it is running out of money before they die.

A recent Barron’s article notes, “for most retirees, the greatest fear is not death—it is running out of money before they die.” As one survey they cite put it, “The prospect of running out of money in retirement is scarier to more people than death.” The same article described a new tool that promises to estimate more accurately how long we are likely to live. Life expectancy is one of the most important assumptions in any retirement plan because it answers a fundamental question: How long will my savings need to last?

The savings need between planning for age 88 and age 100 is enormous. It affects how much you can safely spend each year, how much risk you can take, and how much you should save before retiring.

As we grow older, our perception of our own longevity also changes. We watch friends, siblings, and spouses pass away, and those experiences inevitably shape our expectations. But the real question is not -How long do we think I will live? It is -What happens if we live longer than expected?

Why “we”? Consider a 65-year-old couple. He may have roughly a 35% chance of living to 90; she may have roughly a 45% chance—women consistently outlive men at every age. But the couple’s joint odds are higher still: the probability that at least one of them reaches 90 is significantly greater than either number alone. So the question needs to be what if I or my spouse live longer than expected.

So how should you plan?

Suppose your financial planner’s model estimates that you have only a 20% chance of living past age 88. Does that mean you should confidently spend your last dollar at 88? That does not strike me as prudent.

A well-known economist once posed the issue this way: “Would you accept a 20% chance of running out of money?… Would you play Russian roulette with a six-shooter?”

Planning for a longer life means accepting lower spending throughout retirement. It is the same tradeoff we have faced our entire working lives: enjoy more today or preserve more for tomorrow.

There are few ways to manage longevity risk. The most direct way is creating more guaranteed lifetime income. Delaying Social Security increases your inflation-adjusted lifetime benefit and provides a larger income floor. (Setting aside concerns about the long-term finances of the Social Security system.) Deferred income annuities offer another approach. For example, a 65-year-old might pay $100,000 today in exchange for approximately $5,500 per month beginning 20 years from now if the person is still alive. The payments generally are not inflation-adjusted, but they ensure against the financial consequences of exceptional longevity. And of course one can save more. Each of these strategies requires sacrificing some spending today in exchange for greater financial security tomorrow.

Of course, many retirement plans assume spending naturally declines with age. People travel less, entertain less, and often downsize their homes. Those assumptions are reasonable—but they are not guaranteed.

Just last week, The New York Times profiled someone in their 90s who was still hiking New Hampshire’s White Mountains—I found that challenging in my 50s. A friend recently told me that his mother actually became more mobile in her later years after getting a mobility scooter. Medical advances and technology may extend not only our lifespans but also the number of years we remain active and engaged. If that is true, we may not be able to count on declining health to reduce our spending. And if it does, the costs of that technology may consume the savings. If health and technology extend active years, one might need more money later, not less.

Here is a useful thought experiment. Imagine your longevity model tells you that you have a 20% chance of living past age 87. Now imagine that same model told you there was a 20% chance you would die within the next seven years. Would you spend more today because your life might be shorter? Or would you continue on the same financial because you might live much longer?

We should hesitate to make dramatic changes based on either forecast. That is because retirement planning is not about predicting the future with precision. Planning is about preparing for uncertainty.

My own conclusion is that retirees should generally plan financially for a long life—perhaps age 95 or even 100—while recognizing that discretionary spending will probably decline over time. That approach reduces the risk of outliving one’s assets without assuming one will maintain the same level of spending forever.

While some may want to ‘bounce the check to the undertaker’, that precision in planning is not possible. Longevity forecasts have become more sophisticated. But they should still inform retirement planning, not dictate it. The goal is not to guess the exact year one will die. It is to make sure one’s money lasts if one is fortunate enough to live longer than expected.

I am sure others have thoughts on this difficult issue.

Matt Halperin, CFA, is the founder of Act2 Financial, an app that helps seniors avoid financial fraud. For 30 years, he worked as a portfolio manager and risk manager at large U.S. money managers. Matt currently serves on the investment committee of two endowments.  He has a BA and MBA from the University of Chicago and resides outside of Boston.