Patients with non-muscle-invasive bladder cancer are treated by transurethral resection. About 60-70% of these patients will develop recurrences and in 11% of these cases progression to a muscle-invasive turnout occurs. Surveillance of patients by cystoscopy is therefore carried out every 3-4 months in the first 2 years and yearly thereafter. Several biomarkers have been developed that potentially can detect recurrent bladder cancer in voided urine samples and may present an alternative for the invasive cystoscopy procedure. Recently, van Rhijn reviewed the performance of several of these biomarkers regarding detection of recurrent disease in patients under surveillance. In general, sensitivities were much lower when only patients under surveillance were taken into account than when the patient cohorts included patients with primary disease or patients with high-grade tumours. In this article recent new data on those markers that displayed a sensitivity and specificity of at least 70%, as mentioned in the review by van Rhijn are reviewed. The literature selected was limited to those papers in which the performance of makers was assayed only on urine samples of patients under surveillance. The markers with sensitivity and specificity over 70% that were selected from the previous study are Lewis X, NMP22, microsatellite analysis (MA), CYFRA 21.1, cytokeratin 20 and the UroVysion fluorescence in situ hybridization (FISH) test. Recent new developments such as the use of FGFR3 mutation analysis and methylation detection are also discussed. In conclusion, tests such as the UroVysion FISH test and MA are able to detect most concomitant recurrences and to predict recurrent disease. In general, lesions that are missed are pTa and low grade. With MA several upper tract recurrences were identified that were missed by cystoscopy. The value of the most promising urine tests needs to be established in longitudinal studies and exclusively on patients under surveillance for recurrent disease. A longitudinal setting allows subsequent urine samples to be tested and this increases sensitivity because a negative test outcome sometimes occurs between positive ones. Stratification of patients according to the genetic status of their primary tumours and smoking habits should be investigated. Decision models should be developed that recommend at which points in time cystoscopy or urine testing should be performed.