A new twin study finds that while people tend to choose spouses who are similar to themselves, that mate choice doesn't seem to have any other hereditary component. Twins choose spouses who are no more or less similar to them than non-twins.
The myth that boys tend to choose spouses who are similar to their mothers, while girls tend to choose spouses who are similar to their fathers, also doesn't stand up to empirical scrutiny. People tend to choose spouses who are more similar to them than their opposite sex parent.
The study (from here citing Brendan P. Zietsch, Karin J. H. Verweij, Andrew C. Heath, Nicholas G. Martin. "Variation in Human Mate Choice: Simultaneously Investigating Heritability, Parental Influence, Sexual Imprinting, and Assortative Mating." The American Naturalist, 2011; 177 (5): 605 DOI: 10.1086/659629) involved more than 20,000 subjects and examined traits including height, body mass index, education, income, personality, social attitudes and religiosity.
The only trend discovered, other than a strong tendency of people who choose mates who are like themselves in the measured traits at the time they meet (convergence towards each other in traits over time was ruled out), was a tendency for a woman's family background to influence her choices in terms of age and income.
The tendency to choose someone similar to yourself as a spouse is strongest for age, social attitudes and religiosity, moderate for IQ, education and physical attractiveness, and less strong, but still statistically significant, for height, weight and personality traits.
Pheromone transmitted information about a potential mates MHC complex of genes (related to immunity) influences someone's attractiveness but has essentially no actual impact on who people actually end up in a long term relationship with in the end. The degree of MHC similarity or dissimilarity in married couples is no different that would be expected from random chance.
People also tend to deceive themselves about what kind of person they want in marriage. There is little connection between self-reported preferences in a potential mate and the person that people actually chose to marry.
The study did not attempt to measure which marriages worked and which did not, only what kind of people individuals actually choose to marry in the first place. Thus, the study doesn't rule out the possibility that the best person for someone to marry has traits different than the person an individual is most likely to marry.
In a somewhat related matter, sexual orientation is almost completely unrelated to shared family environment or prevailing societal attitudes. A large (n=7600) twin study in Sweden found that sexual orientation was explained 0%-17% by shared environment (including familial and societal attitudes), 18%-39% by genetics, and 61%-66% by unique environment, "for example, circumstances during pregnancy and childbirth, physical and psychological trauma (e.g., accidents, violence, and disease), peer groups, and sexual experiences." The study was Niklas Långström, Qazi Rahman, Eva Carlström, Paul Lichtenstein. "Genetic and Environmental Effects on Same-sex Sexual Behaviour: A Population Study of Twins in Sweden." Archives of Sexual Behaviour, 7 June 2008 DOI: 10.1007/s10508-008-9386-1. At least in a large share of cases, there is a fair amount of evidence that sexual orientation tends to be congential and is related to some extent to genes and to some extent to hormone exposure in utero and possibly to other epigenetic or in utero enviromental factors.