The associations of work stress, types of work and gender-role orientation with psychological well-being and sickness absence were investigated in a questionnaire survey of 588 male and female nurses and 387 male and female accountants. We hypothesised that health might be impaired among women working in the male-dominated occupation (accountancy), and men in the female-dominated occupation (nursing), but that effects might be moderated by job strain (perceptions of high demand and low control), work and home hassles, and traditional male (instrumentality) and female (expressivity) psychological characteristics. Responses were analysed from 172 female and 61 male nurses, and from 53 female and 81 male commercial accountants. Female accountants were more likely than other groups to have high anxiety scores on the Hospital Anxiety and Depression Scales, while male nurses had the highest rates of sickness absence. Male nurses and female accountants also reported more work-related hassles than did female nurses and male accountants. Men and women in the same occupation did not differ in job strain or job social support, but nurses reported greater job strain than accountants, due to higher ratings of demands and lower skill utilisation. After adjusting for age, sex, occupation, paid work hours and a measure of social desirability bias, risk of elevated anxiety was independently associated with higher job strain, lower job social support, more work hassles, more domestic responsibility, lower instrumentality and higher expressivity. The association between sex and anxiety was no longer significant after instrumentality had been entered into the regression model. Sickness absence of more than three days over the past 12 months was independently associated with higher job strain, more work hassles, lower instrumentality and higher expressivity. The results suggest that when men and women occupy jobs in which they are in the cultural and numerical minority, there may be adverse health effects that are gender-specific. Psychological traits related to socially constructed gender roles may also be relevant, and mediate in part the differences in psychological well-being between men and women.