Lack of reliable prognostic markers hinders accurate prediction of disease progression in prostate cancer. The inducible proinflammatory enzyme cyclooxygenase-2 (COX-2) is implicated in prostate carcinogenesis, but its role in cancer progression is less clear. We examined whether COX-2 expression evaluated by immunohistochemistry (IHC) in radical prostatectomy (RP) specimens can predict biochemical recurrence. Archival prostate cancer specimens (n = 60) were obtained from patients who underwent RP, but had not received neoadjuvant hormonal therapy. Twenty-three patients had biochemical or clinical recurrence (mean time of recurrence: 38.2 months), and 37 patients were recurrence free (mean follow-up: 95 months). COX-2 expression was determined by IHC, using an anti-COX-2 antibody. Three individuals scored the staining independently, as high- or low-expression. COX-2 was expressed in prostate cancer cells, in adjacent normal glands and in specimens from patients who later progressed. At 62-months follow-up, COX-2 staining predicted progression with 82.4% sensitivity and 81.3% specificity. Sensitivity (86.4%) and specificity (86.7%) improved at ≥ 100-months follow-up. In univariate analysis, Gleason score, preoperative PSA, extraprostatic extension, margin, seminal vesicle invasion, and high COX-2 expression were significant predictors of biochemical recurrence (p < 0.05). In multivariate analysis, preoperative PSA (hazard ratio/unit PSA change 1.080; p = 0.0036) and COX-2 expression (hazard ratio 16.442; p < 0.0001) were independent prognostic indicators. Patients with PSA > 7 ng/ml and high COX-2 expression had the highest probability of recurrence (Kaplan-Meier analysis). COX-2 expression is an independent predictor of prostate cancer progression following RP and underscores the significance of inflammatory factors in this process. © 2006 Wiley-Liss, Inc.