OBJECTIVES: Several stimulant and non-stimulant medications are indicated and used alone or in combination to treat attention deficit hyperactivity disorder (ADHD). Little is known about polypharmacy prevalence and predictors in this population. The objective of this analysis was to describe ADHD medication utilization patterns focusing on combination versus monotherapy. METHODS: Health insurance claims from Thomson Reuters MarketScan Commercial Database were analyzed for patients with an ADHD diagnosis (ICD-9 314.0x). Patients included were age ≥6 as of January 2010, continuously enrolled July 2009 through December 2010, and with a 2010 ADHD medication claim. Medication utilization was measured at the patient-month level during 2010. Generalized-estimating equations were used to account for within-patient correlation between months. Demographic and clinical predictors were explored with atomoxetine (ATX), long-acting stimulants (LAS), and alpha-2-adrenergic agents (A2A).  RESULTS: ADHD patients (n=268,172) were primarily age 6-17 years old (57.9%) and male (61.2%). ADHD with hyperactivity was present in 14.2%. Of all treatment-months, 10.3% were combination therapy (more than one ADHD medication class in same month). Short-acting stimulants and A2A agents had the highest combination use (45.3%, 54.0% respectively). ATX, LAS, intermediate-acting stimulants, and prodrug stimulants had the lowest percentage combination use (17.9%, 13.6%, 21.8%, 9.9% respectively), ATX combination therapy was less in older patients (age >18 adjusted odds ratios [AOR] between 0.332-0.829), LAS agents were more frequently used (ages 18-44 AOR 1.079-1.457) and A2A had no association (age >25 AOR 0.699-1.083). Females were less likely to use combination therapy. Hyperactivity was associated with combination therapy in all three models. Tics/Tourettes was associated with combination therapy for ATX and LAS. CONCLUSIONS: Combination therapy rates differed by medication class, as did the demographic and clinical characteristics that significantly predicted combination therapy with ATX, LAS, and A2As.  This suggests these medications may be used differently in clinical practice.