OBJECTIVE: To compare the sensitivity and specificity of UroVysion fluorescence in situ hybridization assay (FISH) with cystoscopy and urine cytology in the surveillance of patients with documented non-muscle invasive bladder cancer (CIS, pTa and pT1). METHODS: This retrospective study was done on a consecutive series of patients undergoing surveillance for non-muscle invasive bladder cancer. The results of FISH were analyzed with concurrent cystoscopy and urine cytology. RESULTS: In all, 94 follow up visits from 59 patients were evaluated. The mean follow up was 52 months. FISH detected 30/48 recurrences of bladder cancer, as compared to 20/48 for cytology and 47/48 on cystoscopy. Hence, the sensitivity of FISH was 63% compared to 42% for cytology (p value 0.03) and 98% for cystoscopy (p value 0.0001). However, cytology was significantly more specific (89%) than FISH (65%) or cystoscopy (41%). FISH was significantly more sensitive in diagnosing Grade 3 tumors (p = 0.0005) than Grades 1 and 2 tumors, when compared with cytology. There was no significant difference in the sensitivity and specificity between FISH and cytology for Grade 1 and 2 tumors. Sensitivity of urine cytology was similar for Grade 3 versus Grades 1 and 2 tumors (p = 0.56). FISH was able to detect all three CIS recurrences whereas cytology was positive in two and atypical in one sample. CONCLUSIONS: FISH has a significantly higher sensitivity than cytology in diagnosing patients with Grade 3 bladder tumors. The low specificity of FISH seen in our study and based on the currently available evidence, the test does not satisfy the criteria for replacing cystoscopy or cytology for surveillance of patients with non-muscle invasive bladder cancer.