Women who breathe wildfire smoke during pregnancy, especially in late stages, may put their offspring at greater risk of autism, a new study of California births suggests.

Researchers found that exposure during the third trimester, when the fetus’ brain grows rapidly, was linked to a higher chance of an autism diagnosis in childhood.

The study — published Jan. 20 in the journal Environmental Science & Technology — looked at health records from more than 200,000 births between 2006 and 2014 in Southern California, a region with both frequent wildfires and higher autism diagnosis rates.

Researchers focused on PM 2.5, tiny particles in wildfire smoke that can enter the lungs and bloodstream. Using home addresses, they estimated how many smoky days pregnant women were exposed to during each trimester.

The strongest link showed up in the last three months of pregnancy, especially when smoke exposure lasted several days in a row.

Compared with no smoke exposure in the third trimester, autism risk was:

• About 10% higher after one to five smoky days

• 12% higher after six to 10 smoky days

• 23% higher after more than 10 smoky days

The association was strongest among women who did not move during pregnancy, suggesting ongoing exposure may matter more than periodical smoke exposure.

Alycia Halladay, chief science officer at the Autism Science Foundation, reviewed the findings.

“This paper supports other scientific research that links prenatal exposure to air pollution, particularly PM 2.5 to autism,” she told NBC News.

“The size of the risk is not huge, but it is consistent with other research and adds to a body of scientific literature linking air pollution and autism," she added.

Senior author Mostafijur Rahman, an assistant professor of environmental health sciences at Tulane University in New Orleans, said wildfire smoke may affect health in a different way than everyday air pollution.

“Wildfire smoke has a unique chemical composition, including higher levels of carbon compounds, metal, toxic byproducts, and it tends to occur in intense and short term spikes,” Rahman said.

Lead author David Luglio, a post-doctoral fellow at Tulane’s Celia Scott Weatherhead School of Public Health and Tropical Medicine, said the timing may be key.

“In terms of the brain, and the late trimester, this is when the brain really grows in size and develops its main centers,” he said.

The study does not prove wildfire smoke causes autism, only that there’s an association. Researchers could not measure how much smoke people inhaled indoors or whether families used air filters, masks or changed behavior during fires.

The effects are small and the pattern isn’t perfectly consistent, David Mandell, a psychiatry professor at the University of Pennsylvania, told NBC News. He said more research is needed before drawing solid conclusions.

Autism affects about 1 in 31 school-aged children in the United States, according to the U.S. Centers for Disease Control and Prevention (CDC). Scientists suspect it owes to a mix of genetic and environmental factors, especially during early life.

Researchers say the findings underscore the importance of reducing smoke exposure during wildfires whenever possible, especially for pregnant women.

“Wildfire smoke is a potentially preventable environmental exposure,” Rahman said.