The weights of biopsy specimens from bovine liver taken for fine needle aspiration cytology (FNAC) by three groups of physicians, were compared. The groups differed from each other in their extent or degree of experience in FNAC. When the physicians used their normal technique, the mean sample weight increased significantly with experience, from 4.6 mg in the least experienced group to 17.2 mg in the group with the longest experience. Thirty-six per cent of the samples taken by the group with the least experience were in the weight class 0-2 mg, whereas no such samples were encountered for the other groups. When all the physicians were advised to use the same, standard technique there were no statistically significant differences in mean sample weight between the groups or in the proportion of samples in class 0-2 mg. Sample weights correlated significantly (P less than 0.001) with cell counts. This study shows that differences in the technique of FNAC are responsible for differences in sample weights and cell counts, including the number of the acellular samples obtained. This could explain in some part the wide variation in the reported accuracy rates of FNAC.