This longitudinal study was originally designed to detect changes in the in vitro immune response of healthy subjects as a result of a psychological intervention. In this study a significant proportion, about 70%, of the immunological variability in the test results was accounted for by the differences in immunological response levels of the subjects. Apart from this between-subject-effect, a significant proportion of the variability in test results was related to the month of data sampling. The month-effect was computed in such a way that the between-subject variation was taken into account. This resulted in a more accurate estimation of the month-effect. Even after correction for the intervention, i.e. the defence of the PhD thesis, the effect of month of data sampling remains significant for mean corpuscular haemoglobin, mean corpuscular haemoglobin concentration, percentage of CD4 and CD8 cells, and for the response to the mitogens phytohaemagglutinin, pokeweed mitogen and concanavalin A as well as the results for the mixed lymphocyte culture for one pool out of three. In contrast, no significant month-effect was observed for the whole blood cell counts, for the differential white blood cell counts as determined by monoclonal antibody staining for cell surface markers CD3, CD16, TAC and OKM1, nor for the immunoglobulin IgM and IgG serum levels. Likewise the cell-mediated lympholysis activities measured against three pools of stimulator cells remained unaltered. We discuss the implications for future immunological follow-up studies of the observation that a significant proportion of the variability in immunological test results is related to differences between subjects and to the month of data sampling.