Divorce in socially monogamous species can result from different mechanisms, for example, chance events, active desertion of the partner, or the intrusion of a third individual ousting the partner. We compared the predictions associated with such mechanisms with data from common guillemots (Uria aalge) breeding on the Isle of May, Scotland. The data cover the years 1982–2005 and show a yearly divorce rate of 10.2%. In most divorces (86%), one of the original partners moved to another breeding site, whereas the other bird stayed and bred with a new partner. On average, movers had a significantly lower breeding success after divorce, stayers were largely unaffected, whereas the incoming birds benefited significantly from the change. This pattern fits best the predictions of the “forced-divorce” hypothesis, suggesting that many divorces were caused by incoming birds rather than the original partners or chance events. Although we are unable to document the precise behavioral sequence that led to divorces, our interpretation is supported by observations of frequent fights over breeding-site ownership. Our data also indicate within-population diversity of divorce mechanisms: some divorces were apparently accidental, others desertion of partners and sites if the latter were of low quality. Our study finally illustrates that a negative correlation between breeding success and probability of divorce (which our data show) need not indicate the adaptiveness of divorce for the original partners. Because such a connection has often been made, adaptive divorce may in general be less common than usually assumed.