OBJECTIVE: Attention deficit hyperactivity disorder (ADHD) affects 3%-5% of North Americans <18 years of age. We conducted a cost-effectiveness (CE) analysis of atomoxetine (ATO), a new non-stimulant, against methylphenidate (MTP) in ADHD. METHODS: Using TreeAge Data 4.0, we constructed a six-month decision analytic model. We included preference rates for non-stimulant over stimulant medication, determined by surveying parents of ADHD children. Efficacy and resource use data were determined from the literature and expert opinion. Perspectives considered were: Ontario Ministry of Health (MoH), which included direct medical costs (physician visits and drug costs); government (MoH and Ontario Ministry of Education), which included MoH costs plus costs of school-based drug administration and special education; and society, including the former costs plus non-pharmacologic interventions such as psychotherapy and over the counter supplements. Prices were obtained from standard lists and measured in 2002 Canadian dollars. Effectiveness was measured as the number of symptom free days (SFDs), defined as days of successful treatment. SFDs were assigned for the 2-week period prior to assessments of successful treatment and for each successful day remaining in the subsequent cycles. Preference rates for ATO were varied in a sensitivity analysis from 57-93% based on 95% confidence intervals of the normalized ATO rate. RESULTS: In the MoH base-case analysis for ATO and MTP, respectively, expected costs were $678 and $162, SFDs were 156 and 98. The incremental cost effectiveness ratio (ICER) was $8.87/additional SFD for MoH perspective, $8.83/additional SFD from the government perspective, and $2.00/additional SFD for societal. In the sensitivity analysis of preference rates, ICERs ranged from $1.12/SFD (societal) to $20.34/SFD (MoH perspective). CONCLUSION: We found a strong preference for non-stimulant therapy, whose incremental cost was relatively small and quite reasonable.