Conventional cervical smears prepared on site by the smear taker are subject to great variation in technical quality and allow little control of the critical parameters required for optimal microscopic diagnosis. More critical, however, is that a significant proportion of the cells removed from the cervix are discarded along with the collecting device and that the material placed on the slide is not transferred in a representative way and may not fully represent the cells removed from the cervix. If the cells were placed directly into preservative fluid, all the material scraped from the cervix would be sent to the laboratory in a well-preserved state, and fully representative slides could be prepared. Several studies have suggested that this would result in an increased cell harvest with a reduction of inadequate slides and an increase in the detection of abnormal cells. Automated preparation devices are now available. Two such devices, CytoRich and ThinPrep, were recently evaluated at the University of Edinburgh Department of Pathology (UEPD). The operational characteristics of each device were evaluated and consumables costed according to 1993-1994 prices. The consumables for the CytoRich were calculated to be more expensive, while operator time for the ThinPrep was more expensive. More mechanical problems were encountered with the ThinPrep, which was also considered more tedious to use. Cytotechnologists required considerable retraining before reaching competence in screening monolayers. They could assess them in approximately half the time required for conventional smears but found it more tiring. Almost no monolayer slides were considered unsatisfactory for laboratory interpretation due to the cells' being obscured by blood, pus or thick streaks of other cells. Scanty monolayers proved particularly difficult to read and interpret. Despite the bias introduced by using only the "leftover" material to make the monolayer, the diagnostic results of the monolayers from both devices were broadly similar to those for the matched split sample conventional smears. UEPD concluded that both devices produced monolayers that were adequate for diagnostic purposes. Neither device was ideal for routine laboratory use in cervical cytopathology in the model tested by UEPD, but both companies recently modified their devices in the light of the deficiencies identified by UEPD. The substantial loss of potentially diagnostic cells during the conventional transfer of cervical scrape material to glass slide alone merits consideration of alternative methods of preparation to obtain representative cell samples for optimal conventional microscopic diagnosis. Monolayers prepared by automated devices offer such an improvement for routine cervical cytopathology. The general deployment of such machines (involving substantial changes in the methods of working of laboratory staff responsible for tens of thousands of clinically important decisions each year) must await the completion of extensive and exhaustive laboratory and field trials. It is recommended that such trials be designed so that all the cellular material removed from the cervix is placed in the cell suspension, and the assessments should be carried out in a routine laboratory environment by cytotechnologists working in a routine cervical cytopathology service.