Up to 30% of cervical intraepithelial grades 2-3 (CIN2-3) lesions regress, but some believe that "regression" is due to "curative" punch biopsies. If this is true, CIN2-3 in the resection margins of the biopsies would be associated with more frequent "persistent" CIN2-3. If, however, immunology-related regression exists, regression would increase with increasing biopsy-cone interval. In 61 punch biopsies diagnosed as CIN3 at careful review by two independent gynaecological pathologists, CIN3 in the resection margins and duration of the biopsy-cone interval was evaluated in relation to CIN2-3-or-not in the cones (again after independent review by expert pathologists). 10 of 61 (16%) patients with CIN3 showed CIN1 or less in the follow-up cones. CIN3-or-not in the resection margins, size of the lesion in the punch biopsy, and presence or absence of CIN2-3 in the cones were not correlated with regression-or-not. However, the number of cones without CIN2-3 increased with longer biopsy-cone interval, 5% in patients with a punch-cone biopsy interval under 9 weeks and 38%> or =9 weeks (p<0.001). These results favour the hypothesis that CIN3 can regress, and do not support the "curative punch biopsy" theory.