In April, 2004, a previously fi t 67-year-old retired surgeon was referred to our clinic with fever and malaise. Between January and March, 2004, he had holidayed in the Canary Islands. His diet had included cheese, fresh vegetables, salads, fi sh, and wine. At the end of his vacation he experienced lumbosacral pain, which was treated by acupuncture when he returned to Germany. On the last day of treatment he developed fever of 40°C, associated with mild nausea, headache, and myalgia. His general practitioner noted fever, absence of leucocytosis, and a high concentration of C-reactive protein (CRP) of 74 mg/L (normal <5 mg/L). He was treated with roxithromycin for 9 days. Ofl oxacin was added when there was no improvement of his symptoms. Since he remained unwell, his general practitioner referred the patient to us. On presentation to our clinic, physical examination was unremarkable. Blood tests showed a slightly high leucocyte count (12·8×109/L), CRP (81 mg/L), and ESR (106 mm/h). Results of liver function tests were abnormal: alanine aminotransferase 112 IU/L (normal <45 IU/L); aspartate aminotransferase 91 IU/L (normal <50 IU/L); γ-glutamyl transpeptidase 67 IU/L (normal <55 IU/L); and alkaline phophatase 137 IU/L (normal <129 IU/L); but bilirubin was normal. The patient’s activated partial thromboplastin time (APTT) was prolonged at 71 s (normal 26–40 s), but prothrombin time and bleeding time were normal. Anticardiolipin antibodies were high (58 kU/L; normal <10 kU/L). Several blood cultures and examination of bone marrow and cerebrospinal fl uid gave no indication of an infectious agent; serological tests for Q fever were also negative. Abdominal ultrasonography, CT of the head, thorax, and abdomen, transoesophageal echocardiography, and MRI of the spine were also unhelpful. Our patient fulfi lled the Petersdorf criteria for fever of unknown origin (FUO). Although his liver enzyme measurements improved during the fi rst week of admission, we thought liver biopsy would be prudent; doughnut granulomas with a fi brin ring were seen (fi gure). A second serological test for Q fever 2 weeks later was positive for phase II antigens of Coxiella burnetii (IgG 470 kIU/L, normal <50 kIU/L; IgM positive; ELISA assay for IgG and IgM, Serion, Wurzburg, Germany). A diagnosis of acute Q fever infection with hepatic involvement was made. Treatment with doxycycline 200 mg daily for 14 days was initiated but the fever persisted. PCR for C burnetii was negative. A reducing dose of prednisolone 40 mg, 20 mg, and 10 mg daily, each for 48 h was introduced. It resulted in a prompt defervescence and resolution of symptoms within 24 h. During the subsequent weeks, the patient’s liver enzyme measurements, APTT, and anticardiolipin antibodies returned to normal, and in August, 2005, he was still doing well. Hepatic involvement of Q fever is not as common as pneumonia; however, in Spain and the Canary Islands, Q fever often results in liver involvement. In our case, the agent was probably contracted by consumption of cheese made from sheep’s milk; hepatic involvement was probably an autoimmune triggered process. The absence of C burnetii on PCR and the presence of the characteristic granulomas in the liver biopsy sample supports the autoimmune hypothesis, although we cannot exclude the possibility that the negative PCR was due to previous antibiotic treatment. This case highlights the importance of considering Q fever in patients with FUO returning from Spain and the Canary Islands. Q fever should not be excluded if initial serology is negative; liver biopsy can be helpful in these patients, and should be considered in those with abnormal liver function. Furthermore, in Q feverhepatitis, steroids should be added to treatment with doxycycline if clinical symptoms do not improve.