The pathophysiology of Epstein-Barr virus (EBV)-associated haemophagocytosis remains poorly understood. 1 In EBVrelated haemophagocytic lymphohistiocytosis, EBV-infected CD8 T cells and natural killer cells are thought to trigger haemophagocytosis and the associated 'cytokine-storm'/systemic infl ammatory response syndrome (SIRS) directly. 2 -4 By contrast, in the context of uncontrolled proliferation of EBV-infected B cells, hyperactively responding EBV-specifi c T cells are assumed to mediate haemophagocytosis/SIRS. 5 Both the quantity and quality of such deregulated EBV-specifi c T-cell reactivity remain undefi ned. Unique insight into basic aspects of the postulated T-cell requirements necessary to trigger haemophagocytosis was provided by the case of a 37-year-old woman with mixed connective tissue disease (ribonucleoprotein-Ab positive, severe pulmonary arterial hypertension, polyarthritis, pericarditis and oesophageal sclerosis). Immunosuppression with azathioprine and ciclosporin, as well as a 3-month trial of oral cyclophosphamide due to progressing pulmonary arterial hypertension was ineffective, yet worsened pre-existing lymphopenia. Based on the severe and cyclophosphamide-resistant course of the disease the indication for haematopoietic stem cell transplantation (HSCT) was established. Before induction-therapy EBV-specifi c MHC I restricted T-cell reactivity was detectable using ELISpot technology, 6 and EBV replication (whole-blood PCR) was controlled ( fi gure 1 ). Subsequent conditioning (cyclophosphamide/ ATG) further depleted CD4 and CD8 T cells and EBV-specifi c T-cell reactivity became undetectable ( fi gure 1 ). Post-HSCT, the patient developed undulant low-grade fevers. Two months post-HSCT she was admitted with fever and coughing. Within 48 h, pulmonary infi ltrations developed, and empiric antimicrobial therapy was initiated. Intriguingly, on the day of hospitalisation, CD8 (and to a lesser extent CD4) T-cell numbers had increased, and strong MHC I and, on this occasion also, MHC II-restricted EBV-specifi c T-cell responses could be elicited ( fi gure 1 ). Indicating immunological ineffi ciency of these responses, however, at the same time increasing EBV viral loads were detected in the blood ( fi gure 1 ), and viral DNA was also found in the bronchial fl uid. Ferritin levels reached greater than 65 000 ng/ml (normal values 10-200 ng/ml). Despite anti-CD20 monoclonal antibody therapy administered on day 4 of hospitalisation multi-organ failure developed, and 1 week after admission the patient died.On pathological examination diffuse alveolar damage and EBVassociated pneumonia were found ( fi gure 2A ). Mononucleosislike EBV-associated lymphoproliferation was observed involving the spleen, lymph nodes, kidneys, liver and basal ganglia. T-cell numbers in the lymph nodes were decreased ( fi gure 2B-D ). In-situ hybridisation identifi ed B-cell blasts expressing EBV-RNA in all of the above locations ( fi gure 2A and B ). No evidence of EBV infection of T cells or natural killer cells was fou...