The normal estimates say that a small fraction of men (perhaps around 3%? 4.5%?) are autogynephilic. The numbers are generally even lower for women with autoandrophilia. I’m somewhat skeptical about this.
First, the autoandrophilia: in my gender surveys, I generally don’t find much lower rates of autoandrophilia among women than autogynephilia among men. There’s a difference in intensity, yes, but this seems to reflect the difference in intensity in sex drives, rather than a difference in frequency. In the past, most transmasculine people seem to have matched the HSTS type, but today there are many queer trans men who very possibly could be autoandrophiles. This makes me think that autoandrophilia is about as common as autogynephilia.
Secondly, I get the impression that the total estimates are too low. They’re often based on transvestic A*P, but this does not seem to be the most common kind of autogynephilia. For example, in my surveys, I generally find twice as many people who get turned on by fantasies about being the opposite sex than by crossdressing. In some social circles, it also seems like there are more people who are out about (likely A*P-caused) trans feelings, and this seems to be more a question of awareness and tolerance of trans issues than of trans people deliberately seeking each other out. (E.g. I have a small circle of friends where several people are out as trans or trans-adj, and this happened long after we met. This seems to be a common experience, and it’s hard to believe that this is the case if A*P is very rare.)
So, what do I think are the real numbers? My surveys generally find rates of around 50%, but that’s too high for me to believe it. As a lower bound, we should probably double the 4.5% number to take more general forms of A*P into account. A plausible upper bound may be a bit higher than 15%; this is the rate that the autopedophilia paper reports when combining two previous samples of straight men.
The idea that the true rate is about 9% to 15% is still a conjecture, and it may turn out that there are flaws with it.