Purpose: To prospectively evaluate the clinical role of urinary NMP22 as a marker for transitional cell carcinoma of the urinary bladder in screening and surveillance settings. Patients and Methods: Single voided specimens were obtained from 211 consecutive patients who presented for flexible cystoscopy. Of these, 96 patients presented with haematuria or irritative symptoms (screening), the remaining 115 were patients with known transitional cell carcinoma on follow–up (surveillance). The urine sample was used for urine microscopy, cytology and for measuring NMP22 levels. Results: Bladder tumours were found in 16 of 96 (16.6%) patients in the screening group and 17 of 115 (15.6%) patients on surveillance. The NMP22 levels were significantly lower in patients with lower stage (Ta vs. T1–3), low grade (G1, G2 vs. G3, CIS) and papillary morphology. The optimum threshold for NMP22 obtained from the ROC curve was 4.75 U/ml, providing a sensitivity, specificity, positive predictive value and negative predictive value of 42.4, 85, 38.5 and 88.6%, respectively. Sensitivity and specificity were better in patients being screened than in those on surveillance. In both groups, urinary NMP22 had similar diagnostic characteristics as urinary cytology. Conclusions: Urinary NMP22 levels are significantly higher in patients with bladder tumour than in those negative for tumours, and test predictability improves with increasing stage and grade. The overall sensitivity for urinary NMP22 is similar to, but not superior to urine cytology. Our study suggests that the clinical role of urinary NMP22 as a diagnostic marker can be at best supportive only.