Recent clinical trials indicated that the ThinPrep method of sample preparation has greater diagnostic sensitivity than the conventional direct Papanicolaou smear. The authors hypothesized that nonhomogeneous cell sampling during transfer from the sampling device to the microscope slide was a contributing factor to the reduced accuracy of the conventional direct Pap smears in these trials. To test this hypothesis, four direct smear methods were compared with the newly developed, fluid-based, filter-transfer method. Counts of epithelial cells on conventional smears showed that only a fraction of the available epithelial cells on the sampling devices (medians, 6.5% to 62.5%) was actually deposited on the slides. In all 27 cases studied with the ThinPrep method, equivalent diagnostic material was obtained on each of the replicate slides prepared per specimen. This identifies a new source of error, preparation error, in conventional smears.