Fluorescence in situ hybridization (FISH) is rapidly emerging as a tool for analyzing numerical and structural chromosomal abnormalities in both liquid and solid tumors. Most studies make use of fresh samples. To determine the feasibility of detecting numerical chromosomal abnormalities (NCA) by FISH using chromosome-specific probes 8, 12, 17, and X (Vysis, Inc., Downers Grove, IL) on archival cytologic preparations, we studied 23 patient samples, one Papanicolaou-and one Diff-Quik-stained slide per case (46 slides), and two additional unstained slides (fresh ascitic fluids) as controls. Included in this study were nine ascitic fluids (four benign and five malignant), four malignant pleural fluids, three benign bladder washes, and seven malignant fine-needle aspirates (FNA) from various sites. The slides ranged from 1-94 days old. After removal of coverslips using xylene, all slides were destained in a series of alcohol and water washes. Pretreatment of slides with pepsin was followed by the in situ hybridization procedure. Two hundred cells per slide were evaluated for distinct separate signals. Results showed the following: 1) all slides were evaluable except for eight (8/46) which had either too few cells or enough cells but with faint signals, 2) the oldest sample showed distinct signals, 3) previously Diff-Quik-stained slides showed relatively better signals than Papanicolaou-stained slides, 4) samples less than a month old showed relatively better signals, and 5) malignant samples showed various NCA, but not the benign samples. We conclude that FISH on archival cytologic preparation 1) is feasible, although age of the slide is a factor since better signals were seen in those less than a month old, 2) shows better results in previously Diff-Quik-stained slides, and 3) is a tool that can be used in the retrospective study of various liquid and solid neoplasms.