The development of a truly non-invasive continuous glucose sensor is an elusive goal. We describe the rise and fall of the Pendra device. In 2000 the company Pendragon Medical introduced a truly non-invasive continuous glucose-monitoring device. This system was supposed to work through so-called impedance spectroscopy. Pendra was Conformité Européenne (CE) approved in May 2003. For a short time the Pendra was available on the Dutch direct-to-consumer market. A post-marketing reliability study was performed in six type 1 diabetes patients. Mean absolute difference between Pendra glucose values and values obtained through self-monitoring of blood glucose was 52%; the Pearson's correlation coefficient was 35.1%; and a Clarke error grid showed 4.3% of the Pendra readings in the potentially dangerous zone E. We argue that the CE certification process for continuous glucose sensors should be made more transparent, and that a consensus on specific requirements for continuous glucose sensors is needed to prevent patient exposure to potentially dangerous situations.