The aim of this study was to evaluate CD8 lymphocyte subsets in active polymyalgia rheumatica (PMR), to determine whether low percentages of CD8+ cells could be used to differentiate PMR from elderly-onset (EORA) and adult rheumatoid arthritis (RA), and to investigate the effects of prednisone on CD8 lymphocyte subsets. A significant reduction of percentages and absolute numbers of CD8bright+ cells was observed in patients with active PMR. Both CD8bright+, CD57- and CD8bright+, CD57+ subsets were significantly reduced. Reduced percentages of CD8+ cells were observed in 55% of patients with active PMR/giant cell arteritis (GCA), in 23% with EORA and in 44% with adult RA. Prednisone therapy in PMR patients, after only 1 week, increased the lymphocyte count and the absolute numbers of lymphocyte subsets significantly. However, the percentages of CD8bright+ cells remained persistently low for the 2 yr study period in 80% of the patients with low pre-treatment levels. Our results demonstrate that CD8 cell percentage is a poor epidemiological discriminator for PMR diagnosis. Notwithstanding the rise in absolute numbers of CD8 cell subsets induced by prednisone, the persistently low percentages of CD8+ cells in a group of PMR patients indicate an abnormality connected with the disease.