There is an excellent discussion of the rescission problem at The Baseline Scenario.
It works like this:
When a person purchases an individual health insurance policy, the company does not bother to verify the information on the application because it costs money to do so. Instead, the company just accepts the premium figuring there is a decent chance the person will turn out to be healthy anyway.
However, if the individual comes down with a major illness, the cost-benefit equation for the insurance company changes. Now it's worth going over the application with a fine tooth comb because the insurance company can save itself a lot of money if it can find some error on the application no matter how inadvertent. For example, a woman who failed to disclose that she was once treated for acne found her policy rescinded when she came down with breast cancer. People have found their policy canceled based on a notation in a chart that they had never seen.
According to health insurance executives, the rescission rate is less than .5% of policies, or 1 in 200, but that number is irrelevant because most people never get their applications examined because they don't have a major illness. If, for example, only 5% of the policies file a major illness claim each year, then the odds of having your insurance policy canceled when you actually need it is 10%. That would mean that 10% of the people who have individual insurance policies should really be considered uninsured because their coverage is an illusion.
I frequently hear conservatives advocate "market based" health care reform plans that involve consumers purchasing the insurance that "is right for them." None of them discuss the rescission problem, however, which flows naturally and inevitably from the profit motive that is the foundation of free market ideology.