Urinary cytology is a non-invasive adjunct to cystoscopy in the diagnosis of bladder cancer. In order to assess the diagnostic accuracy of urinary cytology as an office-based method for clinically relevant high grade (G3) bladder cancer, three nationwide cytology survey tests were performed. Urine specimens from seven patients, three patients with high grade (G3) bladder cancer and four patients with urinary tract infections, were collected. A total of 1,000 cytology slides were produced from each urine specimen. Each set contained five slides (two malignant, three benign) which were sent to all participating German urologists. Three nationwide tests were performed from 1998-2000. The specimen sets were kept the same for the first and second test and in the third test two new slides were introduced. In addition to validity, the reliability was calculated for the first and second test as interobserver and intraobserver reliability according to Cohen's kappa statistics. Due to the change of two specimens in the third test in 2000 only sensitivity and specificity were calculated. A total of 335 urologists took part in the first survey test, 329 in the second and 292 in the third The sensitivity for G3 cytologies was 81.34% in the first, 87.08% in the second and 85.1% in the third survey test and the specificity was 85.87%, 83.58% and 89.15%, respectively. Interobserver reliability showed a weighted kappa value of 0.58 for the first and 0.59 for the second survey test. Calculation of intraobserver reliability was possible for 169 urologists taking part in the first and second survey test and showed a mean kappa value of 0.62. The results of the three nationwide urinary cytology tests indicate that urinary cytology has a high sensitivity in the detection of clinically relevant high grade bladder cancer. The kappa values achieved demonstrate a clear agreement of cytological diagnoses.