More accurately a
value which represents the relative (relative to other people the same age as the subject) ability to do IQ tests.
The values of what someone's mental age should be were never, at least on the Stanford-Binet scale, tested against a real average - thus after World War I there were reports that the mean mental age of US 21 year olds was 13. This was reported in the news as if it showed something wrong with the 21 year olds, when it showed something very wrong with the tests - by their own definition the mean mental age of 21 year olds was 21.
IQ tests were developed by eugenicists with their own racial theories, and skewed, albeit probably unconsciously, towards people of their own class, gender and race, so white, middle-class, Protestant American males still get the highest scores, because much of the test is culturally specific. This would not be a problem were it not that the Stanford-Binet test is the standard against which the others are measured.
The whole notion of an IQ test is fundamentally flawed - 'Intelligence' is defined as 'what this test measures' and then in classic circular reasoning this is used as a proof that the test measures intelligence.
Is Sharon Stone (a member of Mensa) really more intelligent than Richard P. Feynman (IQ measured at 125)?
(some information taken from Stephen Jay Gould's The Mismeasure Of Man
This was updated on 6 March 04 in response to a reply that was nuked even before I'd finished the update. The following is an expansion of my original points, in response to a w/u by a newbie who shall remain nameless
The body mass index is different because it measures something that can be objectively pointed to in the physical world - the relation of body mass to height and build. It is not measuring an undefined, indefinable abstract, but distinct, objectively existing physical properties of an object (in this case, the human body).
Almost any test taken by human beings will provide a normally distributed set of results - that's why it's called a 'normal distribution' , because it is what we would expect any distribution of a random variable to look like. If I were to create a test which tested people's knowledge of the Beach Boys, and test a large enough sample of the population, the result would be a normal curve. Does that mean I'm more intelligent than you because I know Brian Wilson's middle name and the name of Al Jardine's first wife (after all, it must measure intelligence - it's normally distributed!) ? Or does it mean that on any test some people will do well, and some will fail?
If a high IQ is a measure of intelligence, as you claim, then why do you not back up your claim to be in the top 0.000001% of the population with some of the amazingly intelligent things you've done? After all, given that that would mean that you are one of the 60 most intelligent people in the world, you should surely have at least revolutionised physics, cured cancer, discovered some revolutionary new mathematical proof, or at the very least read Everything University and learned how to format text and add links before posting.
Or could it be that you have never done any of these things, and like the majority of people I know who brag about their IQs are a social failure who desperately wants to feel superior to those around him (they're always a 'him') and latches on to the ability to complete a meaningless test as a proof of superiority (see 99% of Mensa members for details of this).
For the record, my own IQ tests at between 170 and 200, depending on factors such as tiredness, which particular version of the test I'm taking and so on, with a median of around 185. It's not through jealousy that I don't accept IQ test results - I would be very happy if despite the massive evidence to the contrary I was vastly more intelligent than the general populace - but because they are simply bad pseudo-science, (and not even the fun kind James Randi gets angry about), and because this pseudo-science actually leads to people's lives getting damaged.
Oh, and if I say I've read a book, I've read the book.