Bumetanide has been reported to alter synaptic excitation-inhibition (E-I) balance by potentiating the action of γ-aminobutyric acid (GABA), thereby attenuating the severity of autism spectrum disorder (ASD) in animal models. However, clinical evidence of its efficacy in young patients with ASD is limited. This was investigated in the present clinical trial of 83 patients, randomised to the bumetanide group (bumetanide treatment, 0.5 mg twice daily) or the control group (no bumetanide treatment). Primary [Children Autism Rating Scale (CARS)], secondary [Clinical Global Impressions (CGI)], and exploratory [inhibitory (γ-aminobutyric acid, GABA) and excitatory (glutamate, Glx) neurotransmitter concentrations measured in the insular cortex (IC) and visual cortex (VC) by magnetic resonance spectroscopy (MRS)] outcome measures were evaluated at baseline and at the 3-month follow-up. Side effects were monitored throughout the treatment course. Compared with the control group, the bumetanide group showed significant reduction in symptom severity, as indicated by both total CARS score and number of items assigned a score ≥ 3. The improvement in clinical symptoms was confirmed by CGI. GABA/Glx ratio in both the IC and VC decreased more rapidly over the 3-month period in the bumetanide group than that in the control group. This decrease in the IC was associated with the symptom improvement in the bumetanide group. Our study confirmed the clinical efficacy of bumetanide on alleviating the core symptoms of ASD in young children and it is the first demonstration that the improvement is associated with reduction in GABA/Glx ratios. This study suggests that the GABA/Glx ratio measured by MRS may provide a neuroimaging biomarker for assessing treatment efficacy for bumetanide.