In a survey of nearly 1,900 U.S. physicians, they found about 30 percent said they'd refer women at average risk of the disease, although several guidelines discourage that.
By contrast, as many as 60 percent said they wouldn't refer a woman at high risk, which guidelines do encourage.
"Physicians aren't doing as good a job as we should at identifying people who should and should not be referred for counseling or testing," said Dr. Michael LeFevre of the U.S. Preventive Services Task Force (USPSTF), who was not involved in the new survey.
Ovarian cancer isn't very common -- it strikes just one in 71 women, many of them elderly, whereas one in eight women get breast cancer.
But because there aren't any good screening tests for ovarian cancer, it's usually not discovered until it's too late.
However, a small percentage of women carry mutations in the BRCA 1 and 2 genes, which make them very likely to develop both breast and ovarian cancer.
Last month, a federal appeals court upheld patents for those two genes by Salt Lake City-based Myriad Genetics.
The company charges $3,340 to test BRCA 1 and 2 for cancer-causing mutations, but told Reuters Health that patients usually end up paying only $100 out-of-pocket.
Still, with just one in 300 or fewer women carrying the mutations, testing those at average risk would put needless strain on the health care system, said Dr. Jacqueline Miller, who worked on the new survey.
"You would be over-testing a lot of women, spending a lot of resources and a lot of money," Miller, of the Centers for Disease Control and Prevention in Atlanta, told Reuters Health.
And it's possible there would be a few false alarms, too, exposing women to unnecessary treatment and other harms, she added.
"For a lot of women, just going through the test creates a lot of anxiety," Miller explained.
The survey, published in the journal Cancer, is based on three fictional patient cases -- one at average risk, one at medium risk and one at high risk.
The USPSTF, a federally supported expert panel, advises against routine counseling and testing for women who don't have suspicious cases of cancer in their family, such as two close relatives with breast cancer, one of whom got it before 50.
"High risk is a little bit complicated," acknowledged LeFevre.
"Physicians can't be expected to carry all of these nuances around in their brain, but I think they should know what the triggers are."
Indeed, the survey suggests doctors who were better at assessing risk were more likely to follow the guidelines.
Yet even when they correctly put a woman in the average risk category, 22 percent of doctors still referred her for counseling or testing.
For those women at high risk, getting genetic counseling and possibly testing may help them decide how they want to deal with that risk.
For instance, 57 percent of women with BRCA 1 mutations get breast cancer by age 70, and 40 percent get ovarian cancer.
Choosing to have the breasts and ovaries removed, or taking certain medications, will cut that risk, said Miller.
"Part of the counseling has to be: here are the options," she explained.
To Miller, the most important lesson from the new finding is to make sure that women at high risk are identified so they can get the right counseling.
But she added that women should never agree to get tested without knowing the reasons.
"You should have that conversation with your provider: why do you feel I'm at high risk?" Miller said.
"If a physician tells you should get genetic counseling, you should understand why."
