Past research into the effects of hardiness on stress and health has yielded inconsistent findings, possibly because of a failure to acknowledge the influence of variables such as negative affectivity and gender. This study examined the main, moderating and mediating effects of hardiness in a sample of 130 (50 male, 80 female) randomly selected university staff members. Controlling for negative affectivity, limited evidence was obtained for the direct effects of hardiness on stress and illness. Kobasa's (1979) model that hardiness buffers the effects of stress on illness was supported for males, but not for females. The hypothesis that approach and avoidance coping mediate the hardiness-illness relationship was not supported for either sex, although there was evidence that relative coping mediated this relationship in females. Collectively, the results point to a need to reconsider the conceptualization and measurement of hardiness.