The current question regarding the reliability of computer-assisted machines demands precise answers in certain highly sensitive application areas, such as emergency medicine or the operation of power plants. This book examines the range of methods developed and proposed in computer science for this purpose. Throughout the investigation, which progresses from specifications and proofs to program testing and the evaluation of various statistical methods, the thesis is reinforced that human understanding—and thus the ability to take individual responsibility—is the success criterion for such methods, for both theoretical and practical reasons.