The present study tested the hypothesis that some subjects may not readily show habituation of adrenocortical stress responses to repeated psychological stress.Twenty healthy male subjects were each exposed five times to the same, brief psychosocial stressor (public speaking and mental arithmetic in front of an audience) with one stress session per day. Salivary cortisol levels were assessed as an index of adrenocortical stress responses. For the total group, cortisol levels were significantly elevated on each of the 5 days. The mean response decreased from day 1 to day 2; however, no further attenuation could be observed on the remaining days. Cluster analysis revealed two groups of subjects who showed completely different response kinetics. In the first group (N = 13), termed "low responders," cortisol levels were elevated on day 1 only. Day 2 to 5 cortisol levels were unaltered. In contrast, subjects in the second group ("high responders") displayed large increases to each of the five experimental treatments. This group had no significant response decrement from day 1 to day 2 to 4 and only a marginal response difference between day 1 and day 5. Discriminant analysis revealed that a combination of five personality scales plus the scores on a symptoms checklist significantly discriminated between high and low responders. With this discriminant function, all 20 subjects were correctly classified to the two groups. These results are discussed with a focus on the possible impact of adrenocortical response types on health and disease.