BackgroundAlthough polygenic risk scores (PRS) predict psychiatric problems, these associations might be attributable to indirect pathways including population stratification, assortative mating, or dynastic effects (mediation via parental environments). The goal of this study was to examine whether PRS‐psychiatric symptom associations were attributable to indirect versus direct pathways.MethodsThe sample consisted of 3,907 dizygotic (DZ) twin pairs. In childhood, their parents rated them on 98 symptoms. In adolescence (n = 2,393 DZ pairs), both the parents and the twins rated themselves on 20 symptoms. We extracted one general and seven specific factors from the childhood data, and one general and three specific factors from the adolescent data. We then regressed each general factor model onto ten psychiatric PRS simultaneously. We first conducted the regressions between individuals (β) and then within DZ twin pairs (β_w), which controls for indirect pathways.ResultsIn childhood, the PRS for ADHD predicted general psychopathology (β = 0.09, 95% CI: [0.06, 0.12]; β_w = 0.07 [0.01, 0.12]). Furthermore, the PRS for ADHD predicted specific inattention (β = 0.04 [0.00, 0.08]; β_w = 0.09 [0.01, 0.17]) and specific hyperactivity (β = 0.07 [0.04, 0.11]; β_w = 0.09 [0.01, 0.16]); the PRS for schizophrenia predicted specific learning (β = 0.08 [0.03, 0.13]; β_w = 0.19 [0.08, 0.30]) and specific inattention problems (β = 0.05 [0.01, 0.09]; β_w = 0.10 [0.02, 0.19]); and the PRS for neuroticism predicted specific anxiety (β = 0.06 [0.02, 0.10]; β_w = 0.06 [0.00, 0.12]). Overall, the PRS‐general factor associations were similar between individuals and within twin pairs, whereas the PRS‐specific factors associations amplified by 84% within pairs.ConclusionsThis implies that PRS‐psychiatric symptom associations did not appear attributable to indirect pathways such as population stratification, assortative mating, or mediation via parental environments. Rather, genetics appeared to directly influence symptomatology.