Mental and physical stress influence the speech process. Mental stress may be, for example, the result of excitement or fear, while physical stress may be caused by fatigue, heat or noise. Speech and speaker recognition systems, designed for non-stressed speech, may collapse when the user is subjected to stress conditions. In this paper we analyze the extent of the problem, will present models of stressed speech and examine means for designing stress-robust recognition systems. We shall also deal with the problem of how to design systems to recognize stress conditions (fear recognition, lie detectors, and so on).