Immunoperoxidase (IP) methods perfected on formalin-fixed, paraffin-embedded tissues (FFPE) and then applied to aspirate smears may result in high background staining and a significant number of false-positive results. This is especially true if polyclonal primary antibodies are used or if aspirates and fluids contain a high interstitial or serum protein content. Because of a recurring problem with antiserum to alpha-fetoprotein (AFP), AFP was selected as the primary test antibody with which to evaluate our avidin-biotin complex (ABC)-IP method. The same method, with diaminobenzidine (DAB) or aminoethylcarbazole (AEC) chromogens, was performed on six types of cytologic preparations of a fresh liver specimen. The liver did not stain for AFP in FFPE and frozen tissue; therefore, it could be used to evaluate potential false-positive staining of direct touch imprints, washed aspirate smears, and cytospins that were both air-dried and alcohol-carbowax-fixed. Initial chromogen incubation times were standardized to give identical results on AFP-positive fixed hepatoma and fetal liver controls. Cytologic preparations immunostained with ABC-IP with AEC chromogen resulted in varying background and hepatocyte staining. In comparison, the ABC-IP method using DAB chromogen resulted in no false-positive results and a clean background. The ABC-IP method with AEC standardized for sensitivity on a fixed tissue control required a markedly shortened chromogen incubation time to preclude significant false-positive staining of cytology specimens. It appears that use of AEC chromogen for this antibody with incubation time standardized on a FFPE tissue control and then applied to cytologic preparations also amplifies nonspecific and background staining, contributing difficulty in assessing a true-positive result.(ABSTRACT TRUNCATED AT 250 WORDS)