Objective. To analyse factors associated with endocervical cone margin involvement and suggest appropriate cone depth for the conization procedure. Design. Retrospective cohort study. Setting. Gynecological oncology center. Population. One thousand two hundred and twenty women undergoing conization for cervical intraepithelial neoplasia (CIN) II or III or stage IA1 microinvasive cervical carcinoma. Methods. The following factors were analysed: age, parity, gravida, conization type, margin status, disease severity and specimen depth. Receiver operating characteristic curve analyses were used to determine the best cut-off points to define appropriate cone depth. Main outcome measures. Cone depth to avoid endecervical margin involvement. Results. Ninety-one women had endocervical margin involvement (7.5%). This was positively associated with disease severity and age and inversely related to cone depth. In women under 50 years of age, the cut-off value was achieved at 1.8 cm cone depth, with high sensitivity and relatively low specificity [area under the curve (AUC) 0.64, sensitivity 0.86, specificity 0.27, p= 0.005]. For a subset of CIN II patients aged ≤50 years, the cut-off value was 1.2 cm (AUC 0.75, sensitivity 0.90, specificity 0.47, p= 0.008). In women <40 years of age, the cut-off value was 1.8 cm (AUC 0.60, sensitivity 0.88, specificity 0.25, p= 0.036). In a subset of CIN II, the cut-off value was 0.9 cm (AUC 0.87, sensitivity 0.83, specificity 0.69, p= 0.002). Conclusions. Age, disease severity and cone depth are predictive factors for endocervical margin involvement. In women of reproductive age, the appropriate cone depth to avoid endocervical margin involvement can be changed depending on age and disease severity.