We conducted a retrospective study to identify the clinical features and surgical observations of congenital cholesteatoma. Sixty patients were diagnosed and underwent surgery for congenital cholesteatoma between April 1987 and May 2002. All diagnoses were made on the basis of two operative findings: 1. the tympanic membrane manifested neither retraction, perforation, nor granulation. 2. the tympanic membrane was not continuous with the cholesteatoma. In this series, congenital cholesteatoma accounted for 7% of all cholesteatomas (853 ears). The patient age ranged from 2 to 48 years. The male to female ratio was 4:1. Seventeen patients had multiple cholesteatoma. Fifty-three patients exhibited closed-type cholesteatomas, while the remaining 7 patients had open-type cholesteatomas that had formed as a flat surface of the epidermis. Patients with open-type cholesteatomas presented with a much more pronounced conductive hearing loss and ossicular erosion or malformation. Twenty-two patients with relatively small cholesteatomas were analyzed to estimate the origin of their cholesteatomas. Of the 22 patients, 13 had anterior superior quadrant (ASQ-type) and 9 had posterior superior quadrant (PSQ-type) cholesteatomas. The mean age at the time of detection was older in the PSQ-type group than in the ASQ-type group and the frequency of ossicular erosion or malformation was more prominent in the PSQ-type group than in the ASQ-type group. The primary site of origin was thought to be the portion between the tympanic ostium of the auditory canal and the semicanal for tensor tympani in the ASQ-type group and near the incudostapedial joint in the PSQ-type group. A planned staged procedure was performed in 29 patients, 15 patients (52%) had residual lesions situated mostly on the oval window, the round window, an exposed facial nerve or an exposed lateral semicircular canal. The frequency of residual lesions in patients who presented with extended, multiple cholesteatoma and those with ossicular malformation was comparable to the frequency of patients who did not present with these features.