Summer abundance of white-footed mice (Peromyscus leucopus) fluctuated periodically during 20 years (1983–2002) at the Holt Research Forest, Maine, USA, a pine-oak forest within 125 km of the northern edge of this species' range. The oscillation period of the series was 4.0 years, with fluctuations greater than in P. leucopus populations in Indiana, Virginia, Ohio, and Pennsylvania. Abundance always increased after a low population phase coupled with a large acorn (Quercus) crop, but in 3 summers population declines followed population peaks despite large acorn crops. We used linear autoregression to identify periodicity after accounting for the relationship between mice and acorn mast. Among 3 candidate models, a 2nd-order lagged abundance (AR[2]) model fit the data better and had greater predictive value (total r² = 0.69) than either a 1st-order (AR[1]) or an acorns-only model. The AR(2) term could represent unknown limiting factors, such as response of predators.