Objective: To investigate the effect of music therapy on eye movement characteristics of joint attention (JA) in children with autism spectrum disorder (ASD), providing an objective basis for clinical intervention. Methods: A randomized crossover design was employed. Twenty-six ASD children aged 3-10 years were enrolled and randomly assigned to group A or B. They first received either a music therapy intervention or a quiet-sitting control condition, and after a 7-day washout period, crossed over to the alternate condition. Eye-tracking technology was used to record the time to first fixation (TFF) during a joint attention paradigm task (comprising three conditions: Eye Gaze, Eye-Head, and Eye-Head-Gesture). Paired sample t-tests were applied to compare the effects of the music intervention. Results: Following music therapy, the TFF for the total score across all conditions was significantly shorter compared to the quiet-sitting control condition (t=-2.24, P=0.035, Cohen's d=-0.458). Specifically, the TFF in the Eye-Head-Gesture condition was significantly shorter than in the control condition (t=-2.32, P=0.030, Cohen's d=-0.485). Although the TFF in the Eye Gaze and Eye-Head conditions alone showed no statistically significant difference, the mean TFF was shorter than in the control condition. Conclusion: Music therapy can significantly shorten the time to first fixation on joint attention targets in children with ASD, suggesting that it may improve core social deficits by enhancing neural information processing efficiency. Eye movement characteristics may serve as an objective indicator for evaluating the efficacy of music therapy.