Microdissection is a feasible tool for the purification of target cells from heterogeneous tissue components. However, the extent to which cells need to be purified by microdissection for use in gene expression analysis has not been determined. In the present study, we obtained diffuse-type gastric cancer tissues at varying purities, and evaluated the corresponding expression of a cancer-specific gene, KRT19, by quantitative real-time PCR. The relationship between the degree of purity and gene expression was confirmed by using 60-mer oligonucleotide microarray analysis. Cancer-specific gene expression was stable in tissues of 10-50% purity, but at 60% or greater purity the slope of the graph was much steeper, indicating a correlation between tissue purity and increased gene expression. Tissues of 70% purity for cancer cells, acquired by microdissection, were therefore deemed to be of sufficient quality to distinguish between gene expression profiles from microdissected and non-microdissected specimens.