The aim of quality control of a laboratory investigation is to ensure that similar results are obtained on the same material at different centres. To investigate its practicability in cytodiagnosis, the same cytological material was examined independently at six centres. Each centre supplied material from 20 cases, providing a total of 120 cases, ie, 100 cases excluding the donor centre's own material. The degree of agreement between the centres was studied using (a) the standard National Health Service cytology report terminology, (b) the centre's own terminology, and (c) the recommended recall time. The results revealed close agreement between five out of six centres in the reports obtained in relation to dysplasia and malignancy, namely, less than 3% false negative results and not more than 1·7% false positive results. The recommended recall time provided a similar order of agreement after discrepancies due to the management of inflammatory conditions had been eliminated. There was marked disagreement in the diagnosis of both presence and type of infection. The results indicate that improvement in the quality of cytological material would increase the consistency of cytodiagnosis. Cytodiagnosis itself, being an expression of opinion, does not appear to be an appropriate field for quality control.