Objective. Neurofeedback can reduce ADHD symptoms; however, current programs are relatively long, with fewer concerns about executive function (EF). The present study aimed to investigate a 20-hour combined computerized training neurofeedback program. Methods. Fifty ADHD children were randomly assigned to either the experimental group (EXP) or the wait-list control group (CON), who took training after the post-tests. The EF measures were the Tower of London (ToL), Wisconsin Card Sorting Test (WCST), and Comprehensive Nonverbal Attention Test (CNAT). SNAP-IV and questionnaires reported by parents constituted the behavioral measures. Two-way repeated-measures ANOVA and bootstrapping dependent t-tests were also used. Results. The F-tests revealed the interaction effects on ADHD symptoms and math scores. The EXP had increased the ToL scores, decreased the error and perseverative error rates on WCST, as well as the dysexecutive index on CNAT in the t-test. Conclusions. The training effects were related to behavioral symptoms and functions, EFs, and generalized achievement performances. We suggest that future studies could apply to different patients and examine the maintenance of the program.