Purpose Urinary nuclear matrix protein (NMP22) was evaluated for detection of new and recurrent bladder tumors in patients with a history of transitional cell carcinoma. Our objective was to determine sensitivity and specificity of this marker for tumors of various stages and grades, as well as its use as an adjunct to or substitute for urinary cytology. Materials and Methods A total of 231 patients with a history of transitional cell carcinoma provided 288 voided urine samples before cytoscopic examination at 1 of 3 institutions (53 patients were reevaluated at least once). Urine samples were assayed for NMP2 using the NMP22 Test Kit. † † Matritech, Newton, Massachusetts. Select patients underwent biopsy with appropriate additional therapy. Voided urinary cytology was obtained in 200 cases. End points for determination of the absence and presence of tumor were negative cystoscopy and positive biopsy, respectively. A receiver operating characteristics curve was constructed to determine the optimal NMP22 threshold for detection of transitional cell carcinoma. For positive biopsies NMP22 values were also correlated with tumor stage and grade. Comparison to cytology was limited to patients with complete data. Results There were 208 negative cytoscopies (158 with cytology) and 66 positive cytoscopies with biopsy (42 with cytology). Of the cases 14 were eliminated from statistical analysis due to incomplete data. Receive operating characteristics curve interpretation determined that 6.4 units per ml. was an optimal reference value or detection of transitional cell carcinoma in this patient group. Sensitivity and specificity for all pathological groupings was 68 and 80%, respectively. When compared to cytology the sensitivities of NMP22 and cytology were 67 versus 31 or 40% (defending on the definition of positive cytology). Conclusion NMP22 values represented significant improvement over urinary cytology for detection of transitional cell carcinoma. The sensitivity of NMP22 for detection of transitional cell carcinoma in bladder cancer patients was as much as twice that of cytology when a reference value of 6.4 units per ml. was used. NMP22 analysis was less costly than cytology and operator independent. While NMP22 has previously been shown to be a strong predictor of recurrence after tumor resection, it is an effective and sensitive screening test for detecting tumors in patients with transitional cell carcinoma.