Here’s the problem: About two-thirds of TDF investors don’t use the funds the way they’re supposed to be used. And their investment performance may be suffering as a result, according to a new study from the independent advisory firm Financial Engines.

A little backstory, first: TDFs, also known as lifecycle funds, invest in a mix of stocks, bonds and cash, based on a specific “retirement target date,” and they shift their asset mix toward bonds and cash as that date nears. The date is usually included in the fund’s name: If you thought you were going to retire in, say, the year 2030, you’d look for a 2030 TDF.

And if your company offers a 401(k), chances are it has a TDF in it. In 2006, Congress gave employers the option to make TDFs the default investment option in their retirement plans, and 72% of such plans now do so, according to research firm Cerulli Associates. Not coincidentally, assets held in TDFs have skyrocketed, rising from around $100 billion in 2006 to more than $800 billion in 2014. More than 40% of retirement-plan accountholders now use a TDF.

But most savers who invest in TDFs don’t stay monogamous, so to speak. Based on Financial Engines’ survey of 1,000 retirement-plan accountholders, only about one in four TDF users keep 90% or more of their retirement-plan assets in their fund. As savers accumulate more money, Financial Engines found, they tend to move some or most of their assets into other funds. About one-third of TDF users have less than 25% of their assets in TDFs. (So much for setting and forgetting.)

And hey, it’s a free country: You should be able to invest wherever you like. But Financial Engines also found that the portfolio of the average “partial-TDF” user underperformed the average investor who kept all or most of her money in a TDF, by a little more than two percentage points a year. That study covered only three years of returns (2010-12), but if such patterns sustained themselves over a longer period, they would add up to a big, painful bite out of a saver’s retirement assets.

Why don’t people stay the course? It’s not an education problem, says Christopher Jones, chief investment officer at Financial Engines: Most investors know that TDFs are supposed to hold all their plan assets. But the company’s interviews with investors suggest that worries about losses make many of them leery of leaning so heavily on one fund—even if that fund is diversified across all kinds of asset classes. “It’s hard to overcome the aversion to having all your eggs in one basket,” Jones tells Fortune.

There’s a Lake Wobegon Effect playing out, too; 61% of “partial” users said they believed they could outperform their plan’s TDFs. “Most investors believe they’re above average,” Jones says.

Jones’s team didn’t attempt to determine why partial users had poorer returns than the “all in” TDF investors. One explanation could be that they’re making their portfolios less diversified—since by buying, say, a U.S. stock fund in addition to their TDF, they’re doubling down on an asset they already own. Some partial users may also be moving their assets among funds fairly frequently, and there’s plenty of academic evidence that, among nonprofessional investors, frequent trading correlates with lower returns.

Jones and Financial Engines recommend that employers combat this problem by encouraging “partial” investors to sign up for managed-account or financial-advisory relationships (services that Financial Engines provides to many 401(k) plans).

But the bigger takeaway is this: Smart asset allocation is key to a successful retirement-investment strategy—and whether you’re making your own allocation decisions, or delegating them to a TDF or an advisor, you’re better off sticking with them. Tinkering around the edges can cost you.