The concept of "gaydar" -- the ability to scan a room and pick out who's gay -- might actually be real.

A controversial new study finds that college students can figure out fairly accurately who's gay or lesbian just by scanning their faces briefly.

And it doesn't come down to what they're wearing or what their hairstyle looks like. The study screened out all those factors by simply presenting faces to the study participants.

The study used 129 college students who were asked to look at 96 black-and-white photos of young adult men that flashed across a computer screen, and then 96 photos of women. Their job was to pick out which faces belonged to people who were gay.

And, just to make sure judgments were not reached based on outfits or hair, the pictures were cropped so that they just showed a floating face. The photos flashed by for only 50 milliseconds -- about a third the time of an eye blink.

The volunteers turned out to be 65 per cent accurate in telling the difference between gay and straight females. Even when the faces were flipped upside down, they were still 61 per cent accurate.

The volunteers had a slightly harder time picking out gay men. Their accuracy with that gender slipped to 57 per cent, and then down to 53 per cent when the pics appeared upside down. But that's still statistically better than the mere 50/50 rate of chance.

Lead author Joshua Tabak, a psychology graduate student at the University of Washington, says "gaydar" might be similar to how we can figure out quickly whether someone is a man or a woman or black or white.

"This information confronts us in everyday life," he said in a statement.

As for why people might be better at picking out lesbian women than gay men, it's possible that the difference between gay and straight women is simply more noticeable in faces, he said.

Tabak notes that the study only quizzed college students. He speculates that "people from older generations or different cultures who may not have grown up knowing they were interacting with gay people" may be less accurate in making gay versus straight judgments.

The study appears in the online journal PLoS ONE.