Objective: We used multi-level modelling of sibling-pair data to disentangle the influence of proband-specific and more general family influences on maternal expressed emotion (MEE) toward children and adolescents with attention-deficit/hyperactivity disorder (ADHD). Method: MEE was measured using the Five Minute Speech Sample (FMSS) for 60 sibling pairs (aged 5 through 17 years) each comprising one proband with ADHD and one child without ADHD. Questionnaire measures were used to assess child and adolescent conduct and emotional problems and maternal depression and ADHD. Multi-level models partitioned the effects of five MEE components (initial statement [IS], relationship [REL], warmth [WAR], critical comments [CC], and positive comments [PC]) into proband-specific and general family effects. Results: Significant proband-specific effects were confirmed for all MEE components, with higher levels of MEE expressed toward probands with ADHD than siblings without ADHD. For REL, PC, and CC, this effect was explained by comorbid child conduct problems rather than ADHD. Only low WAR was associated with child ADHD itself. Furthermore, only low WAR was related to variations in more general family characteristics, especially levels of maternal depression. Conclusions: MEE toward children with ADHD was influenced by proband-specific factors. For most components, these were driven by comorbid symptoms of conduct problems rather than ADHD itself. WAR was different; it was influenced by both child-specific and more general characteristics of the family. Further studies utilising a longitudinal design are required to establish the direction of causation and extend our understanding of the relationship between EE components and ADHD. (Contains 4 tables.)