Most studies of cervical conization have considered the frequency of complications and the outcome of follow-up. The determinants of cone margin positivity have been inadequately described. In a series of CIN patients undergoing conization-equivalent electrosurgical procedure, we evaluated the factors associated with (i) any cone margin involvement, and (ii) endocervical margin involvement (with or without other locations) as contrasted with all other conditions. Study population included 718 patients. Potential determinants of margin involvement were or were treated as categorical. Univariate analysis was based on the chi-square test. Multivariate associations were estimated by multiple logistic regression models. Cone margin involvement was observed in a total of 195 patients (27%). In univariate analysis, the frequency was positively related to histologic grade, time period, lesion size, and cone width and depth. In multivariate analysis, histology diagnosis and time period retained a strong association. The effect of lesion size was of borderline significance. The endocervical location emerged as a multivariate determinant of margin positivity. The effect of cone width and depth was not confirmed. Endocervical margin involvement was observed in 98 cases (14%). In univariate analysis, the frequency was positively associated with histologic grade, time period, and age, and inversely related to the visibility of the squamous-columnar junction. Multivariate analysis confirmed the strong effect of histology diagnosis and time period. The association with age and visibility of the squamous-columnar junction was weaker. Histology diagnosis and time period were the strongest determinants of cone margin involvement. Endocervical margin positivity was also related to patient age and visibility of the squamous-columnar junction. Cone width and depth had no protective effect.