Robust clinical evidence has not been available for melatonin, a drug commonly administered for treating sleep problems of children with autism spectrum disorder (ASD). In a phase 3 randomized, placebo-controlled clinical trial, we administered 1-mg melatonin (n = 65), 4-mg melatonin (n = 65), or placebo (n = 66) to 196 children with ASD once daily before bedtime under adequate sleep hygiene interventions. The primary outcome was sleep onset latency (SOL) assessed with the electronic sleep diary. SOL shortened significantly in the 1- and 4-mg melatonin groups compared to the placebo group (- 22.0, - 28.0, and - 5.0 min, respectively; p < 0.0001 each). This therapeutic regimen of melatonin is a reasonable clinical approach to cope with ASD-emergent difficulties in children with ASD.