Reported urine cytology accuracy, particular sensitivity, is highly variable. We evaluated the accuracy of urinary cytology for primary bladder cancer using population data linkage to provide valid estimates. Consecutive cytology tests processed through a major service between January 2000 and December 2004 were linked to a regional population cancer registry (allowing outcome ascertainment). Sensitivity and specificity were calculated using different thresholds, based on standardized reporting categories (C1 = negative, C2 = reactive, C3 = atypical, C4 = suspicious, C5 = malignant, Cx = inadequate). Cancer registry matching of 2,594 tests revealed 130 incident bladder cancers, of which 97 occurred within 12 months of cytology and were included in calculating accuracy. Sensitivity (C3-C5 considered positive) ranged between 40.2 and 42.3%, and specificity was 93.7-94.1%. If C3 results are counted as negative, sensitivity estimates reduced to 24.7-26.0%. The positive predictive value of a C3, C4 or C5 report was 11.7, 39.2, and 66.6%, respectively. High tumor grade was associated with significantly higher sensitivity compared to low and intermediate grades combined (p = 0.02). Urine cytology is highly specific but has intermediate sensitivity, indicating that it has a role in adjunct diagnosis, but not in screening for primary bladder cancer. C3 results should be considered 'positive' and further investigated, and all positive results should prompt further intervention.