An estimated 145,000 patients will be diagnosed with colorectal cancer in the United States in 1987. Although half of these cancers are potentially detectable by sigmoidoscopy, rigid sigmoidoscopy is not widely used for early detection, largely because of discomfort it causes patients. Flexible sigmoidoscopy has been shown to be more acceptable and more efficient in detecting cancers. In order for flexible sigmoidoscopy to be of more value in cancer control, however, primary care physicians must learn the technique and incorporate it into their complete physical examinations. This paper reports the results of a multicenter trial that evaluated the training required for non-endoscopists to learn how to use the 30-cm flexible sigmoidoscope. Instructions with plastic models, followed by an average of six supervised patient examinations, proved sufficient for them to learn the necessary skills.