I observed razorbill copulation behaviour to examine the ability of males to assure their paternity in the presence of sperm competition. While males in colonial species are generally unable to guard their mates throughout the female's fertile period, male razorbills suffered a special problem in that females actively sought extra-pair copulations (EPCs) and did so in two separate locations, the colony and ledges outside the colony called ''mating arenas''. Most within-pair and extra-pair copulations were performed in the mating arenas (74% and 82%, respectively). Individuals were consistent in their attendance in one of the two arenas, depending upon the accessibility of the breeding site from the arena, and the attendance of other individuals from the same sub-colony. Individuals were also consistent in their occupation of specific sites within the arenas, and usually attended these sites between years. Attendance in the relatively small arenas brought individuals from the sparser colony into much closer proximity than when they attended their nesting sites. The higher density in the arenas was associated with females receiving EPC attempts three times more frequently upon arrival in the arenas than in the colony. Females resisted most EPC attempts, but a high proportion (50%) of females accepted at least one EPC prior to egg-laving. Mating arena attendance by males was aimed at obtaining EPCs when their mates were absent, and at attempting to assure their paternity when their mates were present. Paternity assurance was accomplished by a) inhibiting their mates from accepting EPCs, b), depriving other males access to the female, and c) copulating with the female frequently. Males did not guard their mates effectively in that they were absent from the mating arenas during a large proportion (34%) of their mates' arrivals. The principal male strategy for gaining paternity assurance was apparently to attempt frequent copulations with their mates. Males achieved cloacal contact (and presumably insemination) with their mates approximately 80 times during the 30 days preceding the laving of the single egg, and the number of days males attended the arenas was positively correlated with the number of copulations they performed with their mates. Despite half of females accepting EPCs, EPCs accounted for only 2.4% of inseminations as a result of frequent within-pair copulations.