Work stress, burnout, affective, and physical symptomatology were assessed in 297 hospital nurses. Factor analysis of the scales used to measure these constructs documented their discriminant validity. Structural equations modelling supported the superiority of a model in which burnout functions as an intervening variable between work stress and affective and physical symptomatology, as opposed to one in which burnout is simply another behavioural effect of work stress. A search for a better-fitting model revealed one which shared features with both the initially proposed model in its depiction of burnout as an intervening variable, and with the alternative model, as stress is viewed as having a direct relationship with burnout, affect and physical symptoms. In addition the derived model suggests that physical symptoms are at least partially predicted by affect.