BACKGROUND: Early intervention is effective for reducing ADHD symptoms and related impairments, yet methods of identifying young children in need of services are lacking. Most early predictors of ADHD previously identified are of limited clinical utility. This study examines several theoretically relevant predictors of ADHD in infancy and toddlerhood and whether assessment at multiple time points improves prediction. We also examine whether links between early risk factors and middle childhood ADHD are driven by shared genetic and/or environmental mechanisms to identify potential early intervention targets. METHOD: 614 twins (52.12 % female, 58.50 % non-Hispanic White) participated in a larger study on genetic and environmental contributions to mental health outcomes. Caregiver-reported attention and behavior problems in infancy (M = 12.67 months) and toddlerhood (M = 32.18 months) and caregiver-reported temperament (attentional focusing, impulsivity, inhibitory control) in toddlerhood were used to predict caregiver-reported ADHD symptoms in middle childhood (M = 8.78 years). RESULTS: Only behavior problems in infancy and toddlerhood were robust predictors of caregiver-reported ADHD symptoms in middle childhood. Using multiple timepoints did not improve predictive ability. The relationship between infant behavior problems and caregiver reported middle childhood ADHD symptoms was explained by both environmental and genetic influences, whereas environmental influences almost entirely explained the relationship between toddler behavior problems and caregiver-reported middle childhood ADHD symptoms. IMPLICATIONS: Behavior problems in toddlerhood showed the strongest predictive utility, and links to later ADHD were largely driven by environmental factors, suggesting that psychosocial interventions modifying the child's environment may be particularly effective for reducing ADHD risk.