I recently posted this at Exgay Watch:
Regarding the reasonableness of cross-sex-typed behavior being related to homosexual orientation, I submit:
In a recent meta-analysis of the retrospective literature, Bailey and Zucker (1995)  showed a very strong relationship between extent of childhood cross-gender behavior and a later homosexual sexual orientation for both men and women. However, despite this strong relationship, a proportion of youngsters in Green’s (1987)  follow-up and in the early follow-up results from our clinic (and some other smaller samples) identify as heterosexual. Similarly, in the retrospective literature, not all individuals who later self-identify as homosexual recall a history of cross-gender behavior.
This quote comes from a review article by Bradley and Zucker in the Journal of the American Academy of Child and Adolescent Psychiatry, 1997. The Bailey & Zucker meta-analysis from Dev. Psych, 1995, provides support for the relationship between gender atypical behavior and homosexuality (they predict that 51% of gender atypical boys will become homosexual).
Here is a quote from biological determinist Qazi Rahman allowing that perhaps he could be wrong:
An alternative hypothesis follows the line of Bem’s (1996) developmental model of sexual orientation, and posits an interaction between neuroendocrine prenatal events and postnatal psychosocial influences. Genetic contributors may produce sex-atypical neural differentiation that manifests behaviorally as childhood gender-nonconforming behaviors (which are significantly predictive of adult homosexuality [Bailey and Zucker, 1995]). Differential reinforcements from inputs in the psychosocial milieu to these sex-atypical behaviors makes the “pre-homosexual child” view the same sex as “exotic” (i.e., different from one’s self), which later in puberty becomes the object of eroticization (the “exotic becomes erotic” being Bem’s key theoretical notion). The activational actions of gonadal hormones at puberty may further reinforce this eroticization by fixing key neural substrates in sex-atypical directions, ultimately manifesting in adult homosexual orientation.
To conclude, it is important to illustrate that neurobiological differences between homosexuals and heterosexuals are by no means decisive. Nonetheless, the several independent findings of neuroanatomical differences in sex-atypical directions are not easily refutable. The most plausible explanation for such differences lies in hypotheses implicating sex-atypical neurohormonal differentiation. Unfortunately, evidence currently available is limited and largely correlational in nature. Owing to this, it is not possible for alternative developmental processes associated with sexual orientation to be excluded.
My own belief is that there are people who owe SSA to a Bemian model; some who are youngest sons who as Rahman says, owe the SSA to sex atypical neurohormonal differentiation thus leading to gender atypical phenotype, some people with SSA may experience more social/environmental deficits and so on. None of the evidence explains it all but it all explains something. I am not troubled by alternative pathways to SSA (nor do I think it must be a final destination, ala OSA) because I believe the brain is pretty plastic (retains capacity for differentiation through the lifespan).