Reply #56: Also a good guess, but that's not it either

[Alcibiades]

It's a fairly obvious possible causal variable, but it's not the real variable of interest, and is included in the "we just don't know" message that is the take-home message of the article.

http://www.scientificamerican.com/article.cfm?id=autism-maternal-age

"Maternal age is also increasing in the U.S. A California-based study reported a three-fold increase in the number of births to women aged 40 to 44 between 1982 and 2004. But this trend toward delayed childbearing accounted for less than 5 percent of the total increase in autism diagnoses in California over the decade, according to the study—a finding that surprised Janie Shelton, a doctoral student in University of California, Davis's Department of Public Health Sciences and the study's lead author. "I would have expected to see more of a contribution, because age is a risk factor and women are having kids later," she says."

In other words, 95% of the observed increase in autism-spectrum disorders is not related to maternal age (the article also note that some of the data suggesting that advanced paternal age may be a cause is due to errors in pooling observations in an earlier study). This may be consistent with something environmental, though: older women would have been exposed to whatever environmental factor it might be for a longer period of time, which could account for some of the increased risk.