BACKGROUND: To model and compare contextual (area and family) effects on the psychopathology of children nested in families nested in areas. METHOD: Data from the first two sweeps of the UK's Millennium Cohort Study were used. The final study sample was 9,630 children clustered in 6,052 families clustered in 1,681 Lower-layer Super Output Areas. The mean age of the children at Sweep 2 was 4.96 (SD = 2.76) years. Contextual risk was measured at area level with the Index of Multiple Deprivation (Sweep 1), and at family level with the number of proximal (Sweep 2) and distal (Sweep 1) adverse life events experienced. Psychopathology was measured at Sweep 2 with the Strengths and Difficulties Questionnaire. RESULTS: At baseline, both proximal and distal family risk and area risk predicted broad psychopathology, although the most parsimonious was the proximal family risk model, and both the family-level and the area-level variability were significant. The area risk/broad psychopathology association remained significant even when family risk was controlled, but not when family socioeconomic status was controlled. The full model added parenting and paternal and maternal psychopathology. When parental qualifications were excluded from the family-level contextual controls the effect of area risk remained significant on both externalizing and internalizing psychopathology. CONCLUSIONS: The effect of area on child psychopathology operated via the socioeconomic characteristics of the child's family, not just the adverse characteristics of the neighbors. Multiple family risk predicted child psychopathology directly and independently, and not because it was associated with family socioeconomic status. Family socioeconomic status explained the association between area risk and broad psychopathology.