© 2017 Elsevier Inc. Purpose To examine the effect of celecoxib on cervical intraepithelial neoplasia 3 (CIN 3). This is a NRG Oncology/Gynecologic Oncology Group study with translational biomarkers. Patients and methods Patients with CIN 3 were randomized to celecoxib 400 mg once daily (67 patients) or placebo (63 patients) for 14–18 weeks. The primary outcome measure was histologic regression. A test of equal probabilities of success between two therapies was conducted, using Fisher's Exact Test at alpha = 10% and 90% power when the treatment arm boosted the probability of success by 30%. Translational analysis included cervical tissue HPV genotyping, COX-2 expression in biopsies, and serum celecoxib and VEGF levels. Results In primary analysis, histologic regression was not significantly higher in the celecoxib group (40%) than in the placebo group (34.1%). However, exploratory analyses suggest patients with high serum VEGF levels exhibited greater regression in the celecoxib arm (47.3%) than in the placebo arm (14.3%). Regression rates were similar by treatment group in patients with low VEGF. VEGF levels increased over time in the placebo group, but remained the same in the treatment group. COX-2 expression in cervical biopsies declined from pre-treatment to the end of treatment with celecoxib; it did not change with placebo. Conclusions Celecoxib at 400 mg once daily for 14–18 weeks did not significantly decrease the severity of CIN 3 compared with placebo except, possibly, in subjects with high baseline VEGF. Therefore, serum VEGF levels might identify patients who may benefit from celecoxib or other therapies, personalizing future chemoprevention trials for CIN 3.