OBJECTIVES: To determine when Tropheryma whipplei polymerase chain reaction (PCR) is appropriate in patients evaluated for rheumatological symptoms. METHODS: In a retrospective observational study done in rheumatology units of five hospitals, we assessed the clinical and radiological signs that prompted T. whipplei PCR testing between 2010 and 2014, the proportion of patients diagnosed with Whipple's disease, the number of tests performed and the number of diagnoses according to the number of tests, the patterns of Whipple's disease, and the treatments used. Diagnostic ascertainment was based on 1- Presence of at least one suggestive clinical finding; 2- at least one positive PCR test, and 3- a response to antibiotic therapy described by the physician as dramatic, including normalization of C Reactive Protein. RESULTS: At least one PCR test was performed in each of 267 patients. Rheumatic signs were peripheral arthralgia (n = 239, 89%), peripheral arthritis (n = 173, 65%), and inflammatory back pain (n = 85, 32%). Whipple's disease was diagnosed in 13 patients (4.9%). The more frequently positive tests were saliva and stool. In the centres with no diagnoses of Whipple's disease, arthritis was less common and constitutional symptoms more common. The group with Whipple's disease had a higher proportion of males, older age, and greater frequency of arthritis. The annual incidence ranged across centres from 0 to 3.6/100000 inhabitants. CONCLUSION: Males aged 40-75 years with unexplained intermittent seronegative peripheral polyarthritis, including those without constitutional symptoms, should have T. whipplei PCR tests on saliva, stool and, if possible, joint fluid.