If sexual preferences are acted upon, homosexuality would seem to rule out reproduction, which is most certainly not adaptive in any obviously immediate way.
It doesn't have to be immediate, though.
I like what athon has said so far.
Here's some thoughts:
If homosexuality is influenced by genes, it doesn't mean there is a one to one corelation: "If you have this gene you are gay." it only means it influences it.
It's possible that a certain gene might have the affect in certain environments, or when in a body with other genes, or causing homosexuality. In other environments, or when in a body with other genes, it might not.
On the other hand it might have some other effect.
That's all speculation, but just possibilities.
On the other hand, maybe there is a range of bahaviors likely to come about from homosexuality.
We evolved not in big cities, but as hunter gatherers. As such, that's where we should look at the effects of certain traits. Is it likely that homosexuals didn't reproduce in those societies? I'm not sure, but considering the importance of family on marriage, I do don't think it's unlikely that they did. They may even have been likely to have more offspring that hetrosexuals at times when homosexuals were less common.
Who knows?
It's difficult to look at the exact predicted effects of a trait. Because we can't look at them in a vacuum. Homosexuality may be more than being attracted to the opposite sex (see HeyLeroy's inner ear links, for instance). And the effects of that attraction have to be looked at for the specific environments in which we evolved before we can know what the evolutionary outcome will be.
For instance, here's a scenario. In a certain society, homosexual men, because of societal pressure, are just as likely to marry a woman as are hetrosexual men. But they are less likely to have sexual relationships with other women. As such, they have more time and energy to put into the offspring of their wife. Who might do better because of it than those of hetrosexual men. More survive to maturity to breed on their own.
Of course, if homosexuality spread enough through the population it because easier for homosexual men to have sexual relationships with each other. Thus making them less likely to marry women and have children. Also, those who did have families would now have the option of having extra-marital affairs with other men. Which might decrease their efficiency.
So the frequency of the genes for homosexuality in men would be kept at an equilibrium, a small enough percentage of the population that most homosexual men would be unlikely to come into contact with any other homosexual men, but no less frequent.
This scenario predicts that homosexual men would be as interested in family as hetrosexual men, maybe even more so. Something that at least doesn't seem contradicted by evidence.
That's just idle speculation, but the point is that I can think of ways in which certain environments might be conducive to the evolution of homosexuality, regardless of the fact that it seems at first glance to reduce fitness.