While colonoscopy may detect early-stage colon tumors, a less invasive and more cost-effective technique would be beneficial. Stool, which picks up sloughed-off colonic epithelial cells, would be ideal for sampling the mucosa; shed tumor cells may display alterations in gene expression observed in intact tumors. It is first necessary, however, to show that RNA can be isolated from human feces and that this RNA contains human gene transcripts. We have therefore developed a method for the isolation of total RNA from freshly passed human stool, consisting of lysis in chaotropic agents, repeated extraction with phenol and phenolchloroform, and absorption with an RNA-binding resin. After treatment with RNase-free DNase I, we assayed these preparations for the presence of human RNA by quantitative slot blotting,northern blotting, and reverse transcription-polymerase chain reaction (RT-PCR), We obtained 5-30 mu g RNA per gram of stool from cancer patients, and about 5 mu g RNA per gram of control stool. Quantitative slot blotting showed that about 10% of this RNA was of human origin. Both northern blotting and RT-PCR demonstrated the presence of human RNA in these samples. To unambiguously demonstrate the isolation of RNA from stool, we incubated a mixture of rat cells and control human stool at 37 degrees C for up to 24 hr, RT-PCR of the RNA isolated from this sample clearly revealed the presence of rat-specific mRNA. These experiments indicate that RNA can be isolated from human stool and that message encoded by human genes can be assayed in these preparations. This procedure may provide a powerful tool to identify patients at risk for colon cancer.