Background: Surveillance programmes for bladder cancer are invasive and expensive. Existing guidelines are complex, and the capacity to implement these is untested. The present study examined treatment consistency, and ease of guideline implementation, for patients undergoing surveillance of non-muscle invasive bladder cancer. Method: Eligible cancers treated between 1 January 2005 and 30 June 2009 were identified from a prospective database in a regional South Australian Urology service. Each was analysed with respect to the timing of cystoscopic surveillance and the use of intraoperative chemotherapy. For high-risk patients, the use of urine cytology, upper tract imaging, adjuvant therapy and re-resection of T1 cancers was reviewed. Results: Eight hundred and nineteen cystoscopies were performed in the surveillance of 313 cancers in 193 patients. Within each risk category, the pattern of cystoscopic surveillance varied widely. In high-risk patients, the use of cytology, upper tract imaging, adjuvant therapy and re-resection was infrequent (3-56%). An attempt was made to standardize management through the implementation of guidelines. No overall practice improvement was observed after 18 months. Difficulty incorporating new algorithms into practice and ensuring a consistent longitudinal focus in care were felt contributory. Of 78 low-risk cancer patients, 55% underwent more cystoscopies than would have been expected. In 235 cancer patients at high or intermediate risk, 43% received less follow-up than would have been recommended. Conclusion: Surveillance patterns were inconsistent across all risk categories. The development of consensus recommendations did not significantly alter clinical practice. Implementation of clinical guidelines for this important disease represents a significant challenge in acute hospital settings.