ThinPrep processing, an automated cytopreparatory method, has been reported to show good correlation with conventional preparations and to reduce the rate of false-negative diagnoses. In a retrospective review of 230 consecutive nongynecologic cytology cases, we compare the ThinPrep (TP) method with conventional preparations (CP). There were 129 fine-needle aspiration (FNA) specimens from various sites, including 51 breasts, 40 thyroids, 14 lungs, 8 livers, and 16 miscellaneous sites. The sources of 101 body cavity fluids included 68 pleural/peritoneal effusions, 25 peritoneal/pelvic washings, and 8 miscellaneous sites. Each case was evaluated for cellularity, morphologic, details, and obscuring background material. Diagnoses of the TP slides were classified as insufficient, normal, benign, suspicious, or malignant. Each case was then correlated with the tissue diagnosis when available. In TP slides, cellular arrangements, nuclear details, and nuclear cytoplasmic ratio were preserved, while blood and diathesis were eliminated. There was no statistically significant difference between TP and CP in the diagnostic categories. However, in six cases of "insufficient for diagnosis" on FNA by CP, TP yielded sufficient cells and tissue fragments for diagnosis. One case each of FNA and body fluid with a diagnosis of "suspicious for malignancy" by CP was considered "positive" on TP slides. The overall sensitivity of TP was 97.6%, and the specificity was 92.9%. The positive predictive value was 93.0%. We conclude that the ThinPrep method shows good correlation with conventional preparations in both FNA and body fluids.