It’s common to talk about the turnout challenge facing Donald Trump, who clearly fares well among those who don’t vote regularly. But the candidate with the biggest turnout challenge in this cycle is probably Bernie Sanders.

Sanders appears to be extraordinarily dependent on turnout from infrequent voters, even more than Democrats have recently been in general elections, and maybe more than Barack Obama in the 2008 Iowa caucuses — mainly because his support is so strong among the young.

As Obama can attest, you can turn out and win with irregular voters — and Sanders could prove to have the enthusiasm and organization needed to do the same. In one example, as Jason Horowitz and Yamiche Alcindor reported in The New York Times, the Sanders campaign has plans to send rental cars, vans and buses to carry students who are from Iowa back to their hometowns for caucus day.

But the scale of Sanders’ turnout challenge is unusually large. Compared with the supporters of Hillary Clinton, his are far less likely to report that they intend to vote; they have less history of voting; and they come from demographic categories who turn out in low numbers. This all adds considerable uncertainty to the pre-election polls, which always struggle to determine who is or is not likely to vote.

The challenge for Sanders is most evident in Iowa, a caucus state where polls generally show a fairly tight race.

But the race is not nearly as close among people who say they will definitely vote. The Des Moines Register/Bloomberg poll found Clinton ahead by 9 points among those who said they would “definitely” vote, with Sanders ahead by 10 among those who said they would “probably” caucus. Eight years ago, in the penultimate Des Moines Register poll, Obama had the lead overall and among those who said they would definitely vote.

Similarly, Sanders trails among those who have said they have voted in the caucuses before. A CNN/ORC survey, which showed Sanders up by 8 points overall, gave Clinton a lead of 55 percent to 38 percent among people who reported that they caucused in 2008. NBC/Marist showed Clinton with a 9-point edge among those who indicated past caucus participation.

There are even questions about how many of Sanders’ supporters are actually registered to vote. He has not yet led in an Iowa poll that was conducted using data from voter registration files, the technique preferred by most campaigns but only occasionally used by media pollsters. All but one of the surveys using random digit dialing — a kind of poll that contacts all types of adults, including those who are unregistered — have shown a Sanders lead in the state.

The one poll using random-dialing that showed Clinton ahead was the one from NBC/Marist, and it took the unusual step of matching respondents to the voter file after the survey was completed. That analysis found Clinton ahead by 12 points among people who could be matched to the voter file, while Sanders led by nine points among those who could not be. (It’s possible that the Sanders backers were more difficult to match to the voter file).

It’s important to remember that voters can register on the day of the caucuses, so Sanders could overcome a voter registration problem on Feb. 1, caucus day. But the voter registration statistics from 2008 indicate that only a fraction of voters were unregistered before the caucuses. Turnout increased mainly because of voters who were already registered but had not previously participated in the caucuses.

Sanders’ weakness among “definite” and past voters is underpinned by the crucial demographic divide of the race: age.

There aren’t too many things that affect turnout more than age. Young voters turn out at far lower rates than older voters — even in the elections in which they vote in pretty high numbers. In the 2012 general election, just 45 percent of 18-to-29-year-olds turned out, compared with 72 percent of those over age 65, according to census data.

In lower-turnout elections, these figures are even more lopsided. Not even in the 2008 primary season was Obama able to get the young to vote at near the same rate as older voters.

In the Georgia and North Carolina primaries, where official data on turnout by age is available from 2008, registered black voters who were 18 to 29 were about half as likely to turn out as registered black voters age 65 and older. Similarly, young registered Democrats in North Carolina were only half as likely to vote as older registered Democrats (Georgia does not have party registration).

This is a huge challenge for Sanders: He is competitive because he leads by stupendous margins among younger voters. The NBC/Marist poll showed Sanders ahead by a margin of 64 to 29 among respondents under age 45, but trailing by a 60-to-33-point margin among respondents over age 45. The vaunted Des Moines/Bloomberg poll shows nearly the same breakdown, and so do national surveys.

The split between young and old in the Democratic primary dwarfs the age gap in general elections. It rivals and might even exceed the age gap from the 2008 Democratic primary, whether measured by pre-election or exit polls.

Of course, Obama ultimately won the 2008 Iowa caucuses. He succeeded in mobilizing voters, and his supporters were probably so enthusiastic that he might have earned a strong turnout even without an effective ground operation.

We won’t know whether Sanders repeats his feat until the results come in.