Gallup included the question: "Do you, personally, identify as lesbian, gay, bisexual, or transgender?" for 3 months of surveys, getting 121,000 responses

The always reliable Daily Mail headlines:

'White people are less likely to be gay': Poll reveals African-American community has highest percentage of 'LGBT' adults in U.S. 

Gallup survey, based on interviews with more than 121,000 people, showed that 3.4% of U.S. adults were lesbian, gay, bisexual or transgender (LGBT) 

Highest proportion in black community, at 4.6%, followed by Asians (4.3%), Hispanics (4%) and Caucasians (3.2%)  

Poll found 44% of LGBT adults were Democratic, and 13% Republican

I don't find the results too surprising, but I want to point out a couple of issues: 

There are several percent who don't say know either, just "Don't Know" or "Refuse to Answer."

We don't have any way of knowing what the plain dumb error rate is. (E.g., we can't separate out the male lesbian percentage.) One of the authors of the Gallup Poll, Gary Gates, has discovered in the past that it's not uncommon for a sizable percentage of unusual situations in data to be comprised of typos.

Our work indicates that over 40 percent of same-sex “unmarried partner” couples in the 2000 U.S. Decennial Census are likely misclassified different-sex couples.

There was a flurry of excitement and Brokeback Mountain fantasizing among the, uh, more Andrew Sullivan-like press, over reports after the 2000 Census that there were more gay domestic partnerships in places like Wyoming than in Washington D.C.. But it mostly turned out to be people filling in their Census forms carelessly.

In a study on the military, Gates estimated:

An estimated 2.9% of women on active duty are lesbian/bisexual compared to only 0.6% of men.

Wow, 0.6% of male active duty military personnel are gay / bisexual ...