Taxa are nonarbitrary classes whose existence is an empirical question and not a matter of mere semantic convenience. Taxometric procedures detect whether numerical relations between purported indicators of conjectured taxa bear the hallmarks of true taxa. On the basis of theoretical considerations, the current study tested whether taxa underlie sexual orientation and related measures of gender identity. Two taxometric procedures, maximum covariance, making hits maximum (MAXCOV) and mean above minus below a cut (MAMBAC), were applied to Kinsey Scales and measures of childhood gender nonconformity and adult gender identity in a sample of nearly 5,000 members of the Australian Twin Registry. Results suggest that latent taxa underlie these measures. About 12-15% of men and 5-10% of women belong to latent taxa associated with homosexual preference. These percentages are greater than those of individuals who report homosexual preference, however, and hence it appears that an appreciable proportion of individuals in these taxa have heterosexual preference. An understanding of the origins of these latent taxa may be important to understanding the development of sexual orientation and gender identity.