This study was performed to identify differences in the frequency of specific cytologic diagnoses obtained between specimens prepared by the filtration and ThinPrep (Cytyc, Marlborough, MA) techniques and to assess how these cytologic diagnoses correlated with pathologic findings. Data were collected from 2,347 voided urine specimens analyzed 8 months before and after ThinPrep implementation. Urine cytologic and bladder biopsy specimens were obtained as part of clinical follow-up, and positive diagnoses were considered evidence of malignancy. After ThinPrep implementation, the proportion of specimens diagnosed as negative significantly decreased (85.5% vs 78.6%; P<.001), clusters (5.8% vs 5.6%; P=.597) and positive (2.8% vs 3.3%; P=.143) diagnoses remained similar, and atypical (3.1% vs 8.4%; P<.001) and "suspicious" diagnoses (2.7% vs 4.1%; P<.001) increased. After 1 year of follow-up, there was no significant difference in the percentage of patients diagnosed with bladder carcinoma between the 2 methods for all cytologic categories.