Background: Father absence has negative consequences for children’s behaviour. Yet research has not examined how father absence and child behaviour may influence each other. This study models the cross-lagged relationship between father absence (non-residence) and child problem behaviour in the early years. Methods: We used data from the UK’s Millennium Cohort Study, at children’s ages 3, 5 and 7 years (Sweeps 2-4). The sample was 15 293 families in which both biological parents were co-resident at Sweep 1, when the child was aged 9 months. Child problem behaviour was assessed using the clinical cut-offs of the Strengths and Difficulties Questionnaire (SDQ). We also investigated gender differences in the association between father absence and problem behaviour. Results: Father absence at age 3 predicted a higher probability of the child scoring above cut-off for total difficulties at age 5, as did father absence at age 5 for total difficulties at age 7. There were no significant effects for total difficulties on father absence. Similar father absence effects were found for individual SDQ subscales. Using these subscales, we found few child behaviour effects, mostly during the preschool years: children’s severe externalizing and social (but not emotional) problems were associated with a greater probability of the father being absent in the next sweep. All cross-lagged relationships were similar for boys and girls. Conclusions: Father absence seems to be mainly the cause rather than the outcome of child problem behaviour in young UK families, and to affect boys and girls similarly. There were some child (mostly externalizing) behaviour effects on father absence, particularly in the early years.