To determine the clinical outcome of breast cancer BI-RADS 4 lesions and seek a more effective management guideline, we conducted a retrospective study of all BI-RADS4 patients diagnosed between 2003-2008 with follow up time not less than 2 years. A total of 392 cases of BI-RADS 4 were identified and 320 could be sub-categorised as 4a, 4b and 4c. Overall malignant positive results were 7.65, 38.7 and 58.percent, respectively. In all cases assigned to the close follow up group, no malignancy was detectable (P<0.02). The results of the study suggested that BI-RADS sub-categories have benefit for cancer diagnosis and treatment decisions of clinicians and it might be possible to set up a safe follow-up guideline in selected groups of patients to minimize un-necessary tissue biopsy for breast cancer detection.