Reassociation is the resumption of a romantic relationship between at least one joined Trill and the person who was romantically involved with the symbiont's previous host. Platonic friendships without a romantic element are not subject to this rule, and may be resumed. Since the point of each successive host is for the symbiont to acquire new lifetimes of experiences, Trill consider re-connecting with relationships from a past life "unnatural". Joined Trill who reassociate are exiled from Trill society, and their symbionts die with the current host since they no longer have access to new hosts. This is a dire consequence, since for Trill nothing is more important than protecting the life of the symbiont. (DS9: "Rejoined")
Some things about reassociation seem to conflict with what was seen in TNG: "The Host", when the Odan symbiont appeared willing to rejoin with Beverly Crusher after its transference into a new host. It is possible that this law only applies to two joined Trill. This would make sense considering three of Dax's hosts were friends with Benjamin Sisko, and Ezri Dax's decision to remain with Jadzia's old friends appeared to be perfectly acceptable to Trill society apart from some initial discomfort among the DS9 crew about how to relate to someone who both was and wasn't their old friend. This is also consistent with what René Echevarria said about Michael Piller's original idea for DS9: "Rejoined": "He felt they'd have a very strict taboo in order to avoid an aristocracy of the joined. Otherwise, they'd only want to hang out with each other, their dear old friends from five hundred years ago, and it would become a really screwed up society." (Star Trek: Deep Space Nine Companion): a relation between a joined Trill and a non-Trill (or even an unjoined Trill) would not present this risk.
In the novel Triangle: Imzadi II, Worf speculated that his marriage was over as the new Dax symbiont- even assuming that it was a female- would be automatically discouraged from a further relationship with him due to the rule against reassociation, but this could be the result of his ignorance of the fine details of Trill society where this rule is concerned rather than definite proof that relationships between Trill and non-Trill cannot continue past the life of one host.