This paper reviews experimental contributions published in the last two decades and exploring the effect of emotional stress on neuroendocrine function in healthy humans. Laboratory studies allow standardization of the stressor and better control for known confounding factors. Commonly used stressors are mental arithmetics, speech tasks, the Stroop test, videogame playing, films or videotapes and interviews. Little is known about the generalizability of laboratory results, with some studies suggesting great caution in extrapolating data to real-life stress conditions. Another strategy is studying the psychoendocrine reaction to real-life stressors, such as bereavement or anticipated loss, academic examinations, everyday work and parachute jumping. The effects of different stressors on neuroendocrine axes are reviewed, as well as the influence of gender, age, personality, coping style, social support, biological and nonbiological interventions. The subjective perception of the situation is probably a main determinant of the psychoendocrine response pattern. In fact, marked variability in individual responses to a variety of stressors has frequently been observed. Evidently, the 'objective' characteristics of a given event are not the only determinants of reaction to the event itself. According to a constructivistic perspective, every given stressor has a strictly personal and idiosyncratic meaning and loses its 'objective' characteristics. Of course, biological factors may also play a part. In any case, it is mandatory to overcome a rigid dichotomy between psychological and biological processes. Dualistic conceptions which imply a determination of the physical by the psychological or vice versa should give place to a systemic conception, which implies mutual, circular interactions.