The author outlines the historical attempts made to use infrared thermography for breast cancer detection. He concludes that the technique was a great idea that failed. One can produce good thermal images of the breast, and these images often show features that reflect disease. But the association between skin temperature and breast cancer is too weak to be useful, given the need to detect small tumors and the statistics of breast cancer. However, infrared imaging has experienced great progress in recent years, which is a result of military research in night vision. Several groups in the U.S., Canada, and Japan are experimenting with advanced thermographic systems for detection of breast cancer. This work, the "new thermography", is still in preliminary stages and no extensive clinical studies have apparently been published. The obvious hope is that improved thermal sensitivity will result in improved sensitivity as a medical test -a plausible but so far unproven hypothesis. Detection of breast cancer graphically illustrates the very large difference between getting a medical technology to "work" in a technical or engineering sense, and in meeting the needs of physicians and their patients. There is also a difference between observing promising results with individual patients, and in showing that a technique will benefit the many patients with whom it will be used. We would do our students a favor by helping them to understand the difference. And those who fail to learn the lessons of the past may have to repeat them again.