This retrospective study collected the surgical tissues and the clinical records of 197 surgically treated patients with microinvasive carcinoma of the cervix, which was defined as a limited microinvasion not more than 3 mm in depth regardless of lymph-vascular involvement. Depth of invasion was 1 mm or less in 13.2% and 1.1 to 3 mm in 86.8%. The mean age was 47.2 years. About half of the patients were asymptomatic but suspected cytologically. Random cervical biopsy alone was inadequate for diagnosis. Cone biopsy demonstrated an accuracy of 94% in diagnosis, but failed to eliminate microinvasive carcinoma in 23.0% of the hysterectomy specimens. In 14 cases, the invasion was more extensive in the uterus than in the cone. A positive correlation has been found between the diagnostic rate of conization and the depth of stromal invasion. Cervical conization, while relatively accurate in assessing the existence of stromal invasion that characterizes microinvasion, failed to diagnose those with deeper stromal invasion, especially beyond 1 mm. The recommended approach in diagnosing microinvasion is careful colposcope-directed cone biopsy and proper examination of the conization specimen in order to exclude the possibility of more advanced disease.