Peer behaviour plays an important role in the development of social adjustment, though little is known about its genetic architecture. We conducted a twin study combined with a genome-wide complex trait analysis (GCTA) and a genome-wide screen to characterise genetic influences on problematic peer behaviour during childhood and adolescence. This included a series of longitudinal measures (parent-reported Strengths-and-Difficulties Questionnaire) from a UK population-based birth-cohort (ALSPAC, 4-17 years), and a UK twin sample (TEDS, 4-11 years). Longitudinal twin analysis (TEDS; N = 7,366 twin pairs) showed that peer problems in childhood are heritable (4-11 years, 0.60 < twin-h(2) = 0.71) but genetically heterogeneous from age to age (4-11 years, twin-r(g) = 0.30). GCTA (ALSPAC: N = 5,608, TEDS: N = 2,691) provided furthermore little support for the contribution of measured common genetic variants during childhood (4-12 years, 0.02 < GCTA-h(2)(Meta) = 0.11) though these influences become stronger in adolescence (13-17 years, 0.14 < GCTA-h (2)(ALSPAC) = 0.27). A subsequent cross-sectional genome-wide screen in ALSPAC (N = 6,000) focussed on peer problems with the highest GCTA-heritability (10, 13 and 17 years, 0.0002 < GCTA-P = 0.03). Single variant signals (P = 10(-5)) were followed up in TEDS (N = 2835, 9 and 11 years) and, in search for autism quantitative trait loci, explored within two autism samples (AGRE: N Pedigrees = 793; ACC: N Cases = 1,453/N Controls = 7,070). There was, however, no evidence for association in TEDS and little evidence for an overlap with the autistic continuum. In summary, our findings suggest that problematic peer relationships are heritable but genetically complex and heterogeneous from age to age, with an increase in common measurable genetic variation during adolescence.