P16‐INK4A overexpression has been proposed as a prognostic marker to manage the follow up of women with positive cytology and/or HPV test but without high‐grade cervical intraepithelial neoplasia (CIN2+). This study measures the relative risk (RR) of CIN2+ of p16 positive versus negative in these women. All the women referred to colposcopy from October 2008 to September 2010 with negative or CIN1 colposcopy‐guided biopsy were included in the study; women surgically treated or having a CIN2–3 were excluded. All baseline biopsies were dyed with hematoxylin and eosin and p16. Women were followed up according to screening protocols, with cytology or colposcopy at 6 or 12 months. CIN2/3 RRs and 95% confidence intervals (95%CI) were computed. Of 442 eligible women, 369 (83.5%) had at least one follow‐up episode. At baseline, 113 (30.6%) were CIN1, 248 (67.2%) negative, and 8 (2.2%) inadequate histology; 293 (79.4%) were p16‐negative, 64 (17.3%) p16 positive and 12 (3.2%) not valid. During follow up, we found ten CIN2 and three CIN3; of these, six were p16 positive (sensitivity 46%, 95% CI 19–75). The absolute risk among p16 positives was 9.4/100 compared to 1.7/100 of the p16 negatives (RR 5.5; 95% CI 1.7–17.4). The risk was also higher for CIN1 than for histologically negative women (RR 4.4; 95% CI 1.3–14.3). The RR for p16 in CIN1 did not change (RR 5.2; 95% CI 0.6–47.5). P16 overexpression is a good candidate for modulating follow‐up intensity after a negative colposcopy but is limited by its low prospective sensitivity.