The paper reports recent investigations into mechanisms of insulation failure. After a review of recent progress emphasizing that failure depends on a number of distinct factors which can be separated by suitable techniques, experimental results are given covering some theories recently established. At least five general causes of failure are now clearly distinguishable, and there is strong evidence for two others which are under investigation. The conclusion is drawn that working stresses in insulation can usually be increased when the cause of failure is known in detail and the design modified accordingly. Attention is drawn to the need for close collaboration between laboratory workers investigating causes of failure and design or development staff.