Attention-Deficit/Hyperactivity Disorder (ADHD) is one of the most common neurodevelopment disorders of childhood. Attention, working memory and executive function deficit are considered as core symptoms in cognitive model of ADHD. In recent years, researchers have expanded the cognitive model into neuropsychology and begun to consider the brain abnormality is one of the causes of ADHD. With the development of imaging technology, lots of studies have been conducted to further understand ADHD patients’ brain structural abnormalities, atypical neurodevelopment and dysfunction of brain network connectivity. Researchers used various morphological measurements to find the brain structure differences between ADHD and normal control group. Lots of brain regions have been detected, including fontal, temporal-partial cortex, anterior cingulated cortex, basal ganglia and cerebellum. Results of these studies proved that ADHD has its neural pathology. However, ADHD’s symptoms would change as patients grow up. Some symptoms could disappear as the patient grows into adolescent. Therefore, researchers assumed that ADHD is a kind of neuropsychological development delay disorder. But the clinical and follow-up studies showed that ADHD is multidimensional rather than a simple neurodevelopment disorder. Thus we need to further investigate this disorder from different perspective. Recently, researchers began to treat the human brain as a network which is composed of many nodes (brain areas). They assumed that nodes can communicate and transfer information from each other. Based on this perspective, some researchers hypothesized that ADHD is caused by dysfunction of brain network connectivity. They used lots of statistic methods to test this hypothesis, such as Seed-base Correlation Analysis, Independent Component Analysis, Regional Homogeneity and Graph theory. More and more studies in this areas support this hypothesis, especially the researches of Default Mode Network. There are two controversial views about the dysfunction of DMN in ADHD. One suggests that dysfunction of brain regions connectivity within DMN in resting-state is the main cause of the ADHD. The other view is DMN interference assumption. It suggests that DMN activity is undisturbed at rest but fails to be attenuated during the transition from rest-to-task in ADHD patients. The two views have been supported by different evidences. On the other hand, some researchers used various methods, such as white matter volume measurement and fiber tracking, to study the brain structure connectivity abnormality in ADHD. There are three main findings, including change of the whole brain white matter volume, left fontal-temporal pole and occipital-partial circuit fiber transfer capability. Taken together, these findings suggest that ADHD patients show abnormality in brain functional and structural connectivity. Based on the above three main aspects of ADHD studies, we suggest that behavioral training (such as working memory training) may be a safer way to intervene the disorder or alleviate symptoms than drug treatment. Finally, we concluded four suggestions in future study. First of all, we should classify different kinds of ADHD patients into sub group as each sub-type may link to distinct brain abnormalities. Secondly, it is necessary to combine different kinds of technologies, such as integrating longitudinal method and brain connectivity to figure out how the connectivity patterns change as the patient grows up or the symptoms develop. Thirdly, segment each related brain regions precisely and link them to the ADHD symptoms. At last, we should further compare the effect of drug treatment and behavioral training on the neuropsychological level. Hence, integrating different evidences to understand ADHD’s pathophysiologies may have implications for future anatomy-base differential diagnosis, prevention and intervention.