A b s t r a c t An individually marked population of great reed warblers Acrocephalus arundinaceus was studied between 1984 and 1991 in South Central Sweden. The fates of 279 nests were analysed for the 25 days following clutch initiation. The average frequency of nest loss (43%) did not differ significantly between years or between four periods of the breeding season. After a breeding failure, 70% of the females laid a replacement clutch, most of them together with the same male. On polygnous territories, males assisted the female who first hatched young regardless of her initial mating status (Fig. 2). According to the acutal status of the females attending the nests, for each day nests were classified as being of either monogamous (M), primary (P) or secondary (S) status. From egg-laying to fledging the rate of nest loss decreased among nests of primary status whereas it increased among nests of monogamous and secondary status. During the egg-laying period, the rate of nest loss was 3 times higher among nests of primary than among nests of monogamous and secondary status (Fig. 3). Thus, the high loss level among nests of primary status during the laying period was closely associated with the presence of a female with a less advanced nest on the territory. All nests were situated in reed beds above deep water and most of them at a height at which possible inter-specific nest predators would have caused disturbance to the nest itself. The suspicion that secondary females committed sexually selected infanticide was supported by an experiment with dummy eggs that revealed bill peck markings identical to those obtained from great reed warblers (Fig. 4).