In an analysis of pooled data, we have found that cytotoxic-suppressor T cells outnumber the helper-inducer subset and also the population of cells bearing the pan-T cell marker in specimens of CNS from patients dying with MS (Booss et al. 1983). In the present study of individual data, we have reviewed the case histories to determine if these findings were consistent in various clinical settings. Variables examined included disease duration, tempo of evolution, immunosuppressive therapy, and other potentially immunomodulating features such as tumours. The predominance of the cytotoxic-suppressor subset was not found to be altered by any of these variables. We also present an individual data analysis of cases dying without known CNS disease and of cases with chronic non-inflammatory CNS disease. We found that the low but consistently observed number of T cells was apparently unrelated to the age of the individual or the site of the CNS sampled. Analysis of selected perivascular infiltrates showed, in contrast to the CNS parenchyma, that the pan-T cells and each of the subsets were approximately equal in proportion. Consideration of this observation and of the cytotoxic-suppressor subset preponderance in the parenchyma is compatible with the possibility of antigenic modulation of the T cell differentiation antigens. Finally, the potential contribution of perivascular infiltrates to the CSF pleocytosis is considered.