Multiple fluid specimens of a patient are often received in the cytology laboratory. Both clinicians and pathologists question the optimal number of specimens required to detect a malignancy. We reviewed the computerized cytology files at Boston's Beth Israel Hospital from 1988 to 1991 to identify patients with two or more specimens from the same anatomic site. Two hundred and fifteen patients with a total of 570 specimens were identified. Before December 19, 1990, two direct smears were examined per fluid sample. After December 19, 1990, two direct smears and two cytospin preparations were examined. Medical records of patients without a positive diagnosis of cytology were reviewed. Overall, a cytological diagnosis of malignancy was made on at least one specimen for 55 patients (26%). The first positive diagnosis was made on the initial specimen in 36 patients (65%), on the second in 15 patients (27%), the third in three patients (5%), and the fifth in one patient (2%). For those specimens prepared with the two techniques described above (two direct and two concentrated smears), the first positive diagnosis was made on the initial specimen in 89% of the cases. Medical record review uncovered 55 additional patients who had clinical evidence of malignancy. Of these, 22 (40%) had at least one suspicious diagnosis of their fluid specimens. The first suspicious diagnosis was made with three or fewer specimens in all 22 patients. The majority of malignant effusions are detected with two specimens. Examination of more than three specimens is of little value. Multiple preparatory, especially concentration, techniques may increase the probability of detecting malignancy in one specimen.