We carried out a prospective study to determine the validity of different sets of criteria to define the hypermobility syndrome (HMS), as well as the frequency, reliability and clinical features of HMS items. All consecutive cases of HMS attending the rheumatological outpatient clinic of Hospital del Mar (Barcelona, Spain) constituted the index group (n = 114). A control group of non-HMS rheumatological patients (n = 59) was randomly selected to assess suitable cutoff points and particular HMS item prevalences. Beighton's, Carter's and Rotés' HMS scores correlated very highly among them. Both the correlation coefficients obtained between each pair of sets of HMS criteria and the predictive efficiencies were uniformly high, suggesting high concurrent and predictive validity. All but 2 of the major items were more frequent among women. A basic set of criteria to define HMS is proposed. In relation to previous criteria the new scale shows better internal reliability and homogeneity. Results suggest that it may be suitable for screening studies and in clinical rheumatological settings.