OBJECTIVE Despite the microvascular decompression (MVD) has become a definitive treatment for trigeminal neuralgia (TN) and hemifacial spasm (HFS), not all of the patients have been cured completely so far and this sort of operation is still with risk because of the critical operative area. In order to refine this surgery, we investigated thousands MVDs. METHODS Among 3000 consecutive cases of MVDs have been performed in our department, 2601 were those with typical TN or HFS, who were then enrolled in this investigation. They were retrospectively analyzed with emphasis on the correlation between surgical findings and postoperative outcomes. The differences between TN and HFS cases were compared. The strategy of each surgical process of MVD was addressed. RESULTS Postoperatively, the pain free or spasm cease occurred immediately in 88.3%. The symptoms improved at some degree in 7.2%. The symptoms unimproved at all in 4.5%. Most of those with poor outcome underwent a redo MVD in the following days. Eventually, their symptoms were then improved in 98.7% of the reoperative patients. The majority reason of the failed surgery was that the neurovascular conflict located beyond REZ or the offending veins were missed for TN, while the exact offending artery (arteriole) was missed for HFS as it located far more medially than expected. CONCLUSION A prompt recognition of the conflict site leads to a successful MVD. To facilitate the approach, the craniotomy should be lateral enough to the sigmoid sinus. The whole intracranial nerve root should be examined and veins or arterioles should not be ignored. For TN, all the vessels contacting the nerve should be detached. For HFS, the exposure should be medial enough to the pontomedullary sulcus.