This paper describes the application of the SEM to relatively deep studies of failure mechanisms and of the causes of poor device performance. The first section of the paper outlines ways in which existing SEM's can be integrated with the X-ray microanalyzer, the infrared microscope, the transmission electron microscope, etc., to complement macroscopic electrical measurements. It is shown that limitations arise either because 1) the present instruments are not sufficiently versatile for full exploitation in this type of work, 2) the contrast mechanisms available are not always fully understood, 3) quantitative analyses of the phenomenon exploited are not always available in suitable form, or 4) the irradiation inherent in the SEM alters the device properties. A subsequent section gives a description of work in progress to remove these limitations. In a third section particular problems are examined in depth. These include: 1) the establishment of a rapid, reliable method of obtaining two-dimensional quantitative data. This method is illustrated by measurements on Gunn diodes. 2) The study of dynamic events is outlined with particular reference to thermal problems. 3) The direct observation of thermal hot spots in the SEM is taken as an example of a new contrast mechanism. 4) The methods of eliminating beam-induced changes in devices are outlined.