The accuracy of conization for the prediction of radical hysterectomy (RH) pathological variables in patients with stage Ia2 to Ib1 (≤2 cm) cervical cancer was retrospectively evaluated in the present study. Endocervical or deep resection margin (RM) involvement in the conization specimens was found to be independently associated with residual disease in the hysterectomy specimens (P < 0.001, = 0.003, respectively). When a tumor width of >20 mm in the final RH pathology analysis was predicted by a tumor width of >2 mm or involvement of endocervical or deep RMs in the conization specimens, the sensitivity and negative predictive value (NPV) of conization were 98.2% and 95.2%, respectively. In addition, when deep stromal invasion in the final RH pathology analysis was predicted by deep stromal invasion or involvement of the endocervical or deep RMs in the conization specimens, the sensitivity and NPV of conization were 98.4% and 95.8%, respectively. The sensitivity and NPV of this prediction model for identifying LVSI in the final RH pathology analysis were both 100%. These findings suggest that conization variables and endocervical and deep resection margin statuses can be analyzed to effectively predict RH pathological parameters.