Abnormal saccadic eye movements, such as longer anti-saccade latency and lower pro-saccade accuracy, are common in children with attention-deficit/hyperactivity disorder (ADHD). The present study aimed to investigate the effectiveness of computerized eye-tracking training on improving saccadic eye movements in children with ADHD. Eighteen children with ADHD (mean age = 8.8 years, 10 males) were recruited and assigned to either the experimental (n = 9) or control group (n = 9). The experimental group underwent an accumulated 240 min of eye-tracking training within two weeks, whereas the control group engaged in web game playing for the same amount of time. Saccadic performances were assessed using the anti- and pro-saccade tasks before and after training. Compared to the baseline, only the children who underwent the eye-tracking training showed significant improvements in saccade latency and accuracy in the anti- and pro-saccade tasks, respectively. In contrast, the control group exhibited no significant changes. These preliminary findings support the use of eye-tracking training as a safe non-pharmacological intervention for improving the saccadic eye movements of children with ADHD.