In order to assess whether disclosure of emotions through writing about upsetting or traumatic events resulted in changes in blood-associated immune variables over time, healthy volunteers were randomly assigned to write about either emotional issues or trivial topics for 4 consecutive days. Circulating lymphocytes and T lymphocyte subsets (CD4 (helper), CD8 (cytotoxic/suppressor)) as well as a variety of standard hematological markers were measured following the writing intervention and compared with baseline values. In two separate studies (N 40 and N 38), there were reproducible signi®cant dierences between the emotional disclosure and control writing groups immediately following the intervention in CD4, CD8 and total circulating lymphocyte numbers but not in CD4/CD8 ratios or any other hematological variables. Circulating lymphocyte numbers in the emotional writing group stayed relatively constant over the time-course of the studies, suggesting that the dierence between the groups was due to a transient elevation in postwriting blood lymphocyte numbers in the control group. Self-evaluations immediately before and after each writing session con®rmed that the intervention was stressful for subjects in the emotional disclosure group but the eects on circulating lymphocytes were not attributable purely to anxiety-related factors. The results extend our previous observation of changes in immune reactivity following a writing intervention and indicate that the group dierences are the result of ¯uctuations over time in the control group but relative stability in the emotional disclosure group. It is conceivable that such `buering' of temporal immune variation might be in¯uential in the health-promoting eects of emotional disclosure.