To estimate the prevalence of FMS and to examine the clinically derived concept of FMS in a population sample. Methods: Five hundred forty-one German residents of Bad Sackingen, aged 25-74, were randomly selected for study from a target population of 6,100. Four hundred thirty-eight [81%] responded to a postal screening questionnaire. One hundred five of these reported chronic pain in at least one joint region and in the axial region [i.e., neck, back and/or thoracic pain]. All were invited to a medical examination: 80 attended. Cases of FMS were identified by counting active tender points [TP, n=34] and non tender control points [CP, n=10]. A proband was classified as FMS if she/he showed 17 or more active TPs and 2 or fewer tender CPs. Results: Eighteen subjects had 17 or more TPs. Eight of these had 3 or more tender Cps resulting in 10 FMS-cases and a minimal prevalence of 1.8% The median age was 62; 8/10 were females. Assuming that the prevalence in the nonresponders and nonparticipants was similar to those who responded and participated, and that none of those with regional or no musculoskeletal pain fulfilled the TP/CP-criteria, the estimated prevalence would be 3.1% [95% CI 1.6-4.4]. Using the ACR-criteria (Wolfe et al. 1990) results in a rate of 2.0. The second assumption however may not be valid: in a small control group of 20 probands with only monolocular pain but a high amoutn of non-specific bodily complaints, 3 were found to fulfill our TP/CP-criteria for FMS. Conclusions: Our study, though giving a prevalence estimate close to that of the other FMS-surveys, raises questions about the rheumatologically defined nosologic and nosographic concept of the disorder.