In the surgical development of cholesteatoma, in order to reduce the recurrence of cholesteatoma, 2 kinds of surgeries were carried out: removal of Cog and Korner's septum to ventilate supratubal recess (supratubal recess opened) and obliteration of the mastoid and attic space (bony obliteration tympanoplasty) were invented, respectively. Their purpose is the same but the theoretical basis is different, and the comparison of these 2 methods is not reported in the current literature. This study aims to evaluate the rates of recurrent and residual cholesteatoma with the simple canal wall up and canal wall up-supratubal recess opened and canal wall up-bony obliteration tympanoplasty mastoidectomy in a large cohort of patients. The secondary objectives were to assess the 3 techniques' infection rates and hearing outcomes. Overall, 352 patients with middle ear cholesteatoma preoperatively underwent temporal bone ultrahigh-resolution computed tomography scan. The shape of the Eustachian tube and the supratubal recess were analyzed, and superior and posterior tympanic recesses, including the supratubal recess, were opened in different surgical groups. After 5 years of follow-up, the results show that the lowest recurrence rate was 6.6% (7/106) for canal wall up-supratubal recess opened, 10.9% (12/101) for canal wall up-bony obliteration tympanoplasty, and canal wall up had the highest recurrence rate of 19.31% (28/145). The postoperative infection rate was 5.7% in the canal wall up-supratubal recess opened group, 10.89% in the canal wall up-bony obliteration tympanoplasty group, and 7.59% in the simple canal wall up group. The postoperative median air conduction was increased 8 dB in the canal wall up-supratubal recess opened group, 1 dB in the canal wall up-bony obliteration tympanoplasty, and 6 dB in the simple canal wall up group. Opening the supratubal recess to ensure the patency of the attic facilitates the gas exchange between the mastoid process and the middle ear and reduces the possibility of cholesteatoma recurrence.