Objective: To evaluate the hypothesis of stimulant medication effect on physical growth in the follow-up phase of the Multimodal Treatment Study of Children With ADHD. Method: Naturalistic subgroups were established based on patterns of treatment with stimulant medication at baseline, 14-, 24-, and 36-month assessments: not medicated (n = 65), newly medicated (n = 88), consistently medicated (n = 70), and inconsistently medicated (n = 147). Analysis of variance was used to evaluate effects of subgroup and assessment time on measures of relative size (z scores) obtained from growth norms. Results: The subgroup x assessment time interaction was significant for z height (p less than 0.005) and z weight (p less than 0.0001), due primarily to divergence of the newly medicated and the not medicated subgroups. These initially stimulant-naive subgroups had z scores significantly greater than 0 at baseline. The newly medicated subgroup showed decreases in relative size that reached asymptotes by the 36-month assessment, when this group showed average growth of 2.0 cm and 2.7 kg less than the not medicated subgroup, which showed slight increases in relative size. Conclusions: Stimulant-naive school-age children with Combined type attention-deficit/hyperactivity disorder were, as a group, larger than expected from norms before treatment but show stimulant-related decreases in growth rates after initiation of treatment, which appeared to reach asymptotes within 3 years without evidence of growth rebound. (Contains 3 tables and 2 figures.)