In this paper we study marriage formation through a two-sided secretary problem approach. We consider individuals with nontransferable utility and two different dimensions of heterogeneity, a characteristic evaluated according to the idiosyncratic preferences of potential partners, and an universally-rankable characteristic. There are two possible states of the world, one in which people meet their partner randomly, and one in which the meeting occurs between individuals with similar characteristics. We show that individuals with higher universal characteristic tend to be more picky in their marriage hunting. This does not necessarily mean that they marry later than other individuals, since the higher expected quality of their potential partners in the assortative meeting state can make them marry earlier than individuals with a lower universal characteristic.