Studies on marriage and divorce often assume, explicitly or implicitly, that there is a positive relationship between partner similarity and well-being. We test this similarity hypothesis: do individuals who share more socio-demographic characteristics with their partners report higher well-being than individuals whose partners are less similar? We analyzed information on more than 2,300 married and cohabiting couples aged 18-50 from the UK Understanding Society wave 1 survey. Three dimensions of well-being were assessed: relationship quality, life satisfaction and psychological distress. We examined similarity on seven characteristics separately and as an index of similarity: age, father’s class, education, ethnicity, religiosity, native language, and parental divorce. The results provided no support for the similarity hypothesis: there was no evidence for a positive association between partner similarity and the three well-being measures. We discuss the implications of this finding for our understanding of partner choice and divorce.