Objectives: To use growth mixture modelling (GMM) to identify subgroups of children with attention deficit hyperactive disorder (ADHD) who have different pharmacodynamic profiles in response to extended release methylphenidate as assessed in a laboratory classroom setting. Methods: GMM analysis was performed on data from the COMACS study (Comparison of Methylphenidates in the Analog Classroom Setting): a large (n = 184) placebo-controlled cross-over study comparing three treatment conditions in the Laboratory School Protocol (with a 1.5-h cycle of attention and deportment assessments). Two orally administered, once-daily methylphenidate (MPH) bioequivalent formulations [Metadate CD™/Equasym™ XL (MCD-EQXL) and Concerta XL (CON)] were compared with placebo (PLA). Results: Three classes of children with distinct severity profiles in the PLA condition were identified. For both MCD-EQXL and CON, the more severe their PLA symptoms the better, the children's response. However, the formulations produced different growth curves by class, with CON having essentially a flat profile for all three classes (i.e. no effect of PLA severity) and MCD-EQXL showing a marked decline in symptoms immediately post-dosing in the two most severe classes compared with the least severe. Comparison of daily doses matched for immediate-release (IR) components accounted for this difference. Conclusion: The results suggest considerable heterogeneity in the pharmacodynamics of MPH response by children with ADHD. When treatment response for near-equal, bioequivalent daily doses the two formulations was compared, marked differences were seen for children in the most severe classes with a strong curvilinear trajectory for MCD-EQXL related to the greater IR component. © 2007 Steinkopff Verlag.