A plague of mice (Mus domesticus) in the Victorian mallee wheatlands of south-eastern Australia in autumn 1984 appeared to be generated by a sequence of rainfall events: high autumn (March), mid winter and late winter rainfall in 1983, and high summer rainfall in 1983/84. The March rainfall in 1983 ended a drought; mice began to breed and bred until the end of May. Relatively high survival of mice for 12 months after March 1983, together with early onset of breeding and high reproductive performance throughout the 1983/84 breeding season, including summer, were key demographic processes during the formation of the plague. Temporal differences in mouse abundance and breeding performance between habitats highlighted the relevance of specific habitats to the dynamics of mouse populations in the wheatlands. Fencelines were the most important habitat of mice because they were foci for breeding at the start of the breeding season, good nesting sites which were rarely disturbed, and widespread and in close proximity to crops. Cereal crops were colonized in spring 1983 and in autumn 1984; they became important habitats in 1983 when mice dispersed and bred there in early spring. Redhead's (1988) model was sufficient to explain the 1984 plague, but not the magnitude of the decline of mouse numbers in 1984, nor the absence of a further outbreak in 1985. A new model is proposed based on a sequence of rainfall events beginning at least 10 months prior to a plague.