A major problem in the management of SLE patients is to predict a flare or to distinguish between active and quiescent disease. Serological markers are widely used to assess disease activity, but many patients have close to or normal values for these parameters while exhibiting obvious disease-related signs and symptoms. This study aimed to determine which serological parameters, among ESR, ANA and anti-dsDNA antibody titres, CH50 and the HLA-DR expression on circulating T-lymphocyte subsets, best reflected the development of SLE flares. Sixty SLE patients were included, 34 with quiescent disease throughout the entire follow-up period and 26 who experienced an SLE flare defined as having active disease. According to univariate analysis, all parameters were significantly higher for patients with active disease, with the percentage of CD8+DR+ cells being the most significant parameter (P = 10-7). Multivariate logistic regression analysis identified three independent variables enabling the identification of a lupus flare: CH50, the CD8+DR+ and CD4+DR+ cell percentages among total lymphocytes. The CD8+DR+ cell percentage is the biological parameter most significantly associated with a flare (P < 0.001), even more powerful than CH50 (P < 0.01). HLA-DR expression on CD8+ lymphocytes clearly coincided with disease evolution in seven patients enrolled as having quiescent disease, but who experienced one flare during follow-up that subsequently resolved. The percentage of circulating CD8+DR+ lymphocytes appears to be a biological marker which accurately reflects disease activity. A larger prospective study is needed to demonstrate the real efficacy of this marker in predicting an exacerbation in SLE patients.