Weasels are usually thought of as ground predators, subsisting largely on mice and voles, but they are equally adept at climbing trees to plunder the nests of birds. With the widespread introduction of nest boxes throughout Europe for bird population studies, the weasel soon emerged as a regular and sometimes highly destructive predator of eggs, young, and occasionally the parent birds (Perrins 1965; Graczyk, Galinski & Klejnotowski 1970; Larsen 1974; Flegg & Cox 1975). An interesting feature common to the long continuing studies is that predation tends to fluctuate widely from one year to the next, but usually the investigation has been too brief to establish the reasons for this. In Wytham Wood near Oxford, data on the breeding success of four Parus species of tits in next boxes have been collected continuously since 1947, and the causes of nest failure, including predation by weasels, have been carefully recorded. Long term records of rodent abundance are also available for Wytham (Southern & Lowe 1968; Southern 1970); these monitored populations of Bank Voles Clethrionomys glareolus (Schr.) and Wood Mice Apodemus sylvaticus (L.) which are the commonest small mammals in the wood and the main prey species of the weasels there, comprise respectively 36 and 14% of their diet (King 1971, 1975). A k-factor analysis by Krebs (1970) showed that in years of high nest density, predation on Great Tits Parus major L. was proportionately more severe; a detailed examination of the fate of nests in relation to the distances between occupied nest boxes showed that, within the range measured, there was no threshold density below which the birds were completely safe (Krebs 1971). The present study is likewise confined to the breeding season of the tits which occupies the three months from April to June. Using data mainly from 1947-72, the first aim of the study is to assess the intensity of the weasels' predation on tits in relation to the abundance of both tits and rodents. The second objective is to examine differences in the timing of predation within the breeding season. Perrins (1965) found consistent differences between the risk of predation to early and late layers, and attributed this variation to a change in the behaviour of both prey and predator. This paper explores the possibility that differences in the timing of predation may also arise from variations in prey density.