Gamma aminobutyric acid (GABA)-mediated synapses and the oscillations they orchestrate are altered in autism. GABA-acting benzodiazepines exert in some patients with autism paradoxical effects, raising the possibility that like in epilepsies, GABA excites neurons because of elevated intracellular concentrations of chloride. Following a successful pilot study,1 we have now performed a double-blind clinical trial using the diuretic, chloride-importer antagonist bumetanide that reduces intracellular chloride reinforcing GABAergic inhibition. Sixty children with autism or Asperger syndrome (3–11 years old) received for 3 months placebo or bumetanide (1 mg daily), followed by 1-month wash out. Determination of the severity of autism was made with video films at day 0 (D0) and D90 by blind, independent evaluators. Bumetanide reduced significantly the Childhood Autism Rating Scale (CARS) (D90−D0; P<0.004 treated vs placebo), Clinical Global Impressions (P<0.017 treated vs placebo) and Autism Diagnostic Observation Schedule values when the most severe cases (CARS values above the mean±s.d.; n=9) were removed (Wilcoxon test: P-value=0.031; Student's t-test: P-value=0.017). Side effects were restricted to an occasional mild hypokalaemia (3.0–3.5 mM l−1 K+) that was treated with supplemental potassium. In a companion study, chronic bumetanide treatment significantly improved accuracy in facial emotional labelling, and increased brain activation in areas involved in social and emotional perception (Hadjikhani et al., submitted). Therefore, bumetanide is a promising novel therapeutic agent to treat autism. Larger trials are warranted to better determine the population best suited for this treatment.