Objective: To explore placebo-controlled efficacy and safety of atomoxetine (ATX) for attention-deficit/hyperactivity disorder (ADHD) symptoms in children with autism spectrum disorders (ASD). Method: Children ages 5 to 15 with ASD and prominent ADHD symptoms were randomly assigned to order in a crossover of clinically titrated ATX and placebo, 6 weeks each, separated by 1-week washout. Slopes for each condition were compared by paired t test. Results: In 2004-2005, 12 boys and 4 girls (7 with autistic disorder, 1 Asperger's, 8 pervasive developmental disorder not otherwise specified) all completed at least 3 weeks of each condition. On the primary outcome, the Hyperactivity subscale of the Aberrant Behavior Checklist, ATX was superior to placebo (p = 0.043, effect size d= 0.90). It was also superior on a 0 to 3 rating of nine DSM-IV ADHD hyperactive/impulsive symptoms (p = 0.005, d = 1.27), but missed significance on nine inattentive symptoms (p = 0.053, d = 0.89). Nine subjects responded to ATX, four to placebo (25% improvement on the Hyperactivity subscale plus Clinical Global Impressions-Improvement of 1-2. One was rehospitalized for recurrent violence on ATX. Adverse events were otherwise tolerable, with no tendency to stereotypy. Conclusions: ATX appears safe and effective for treating hyperactivity in some children with autism spectrum disorders. The effect appears as large as in a multisite methylphenidate trial in the same population, with fewer intolerable side effects. Further study in autism spectrum disorders is indicated. (Contains 1 figure and 4 tables.)