The aim of this study was threefold: (1) to investigate the prevalence of restless legs syndrome (RLS), in Ankara, Turkey; (2) to determine the predictive values of diagnostic criteria; and (3) to determine the frequency of physician referrals and the frequency of getting the correct diagnosis. A total of 815 individuals, from randomly selected addresses, above the age of 15, were reached using the questionnaire composed of the four diagnostic criteria. Individuals who responded by answering 'yes' for at least one question were interviewed by neurologists for the diagnosis of RLS. Frequency of physician referrals and frequency of getting the correct diagnosis of RLS were also determined for patients getting the final diagnoses of RLS. Prevalence of RLS in Ankara was 5.52 %; 41.0 % of the individuals diagnosed with RLS had replied 'yes' to either one, two or three questions asked by interviewers. However, only 21.3 % of individuals who replied 'yes' to all four questions received the diagnosis of RLS. Among the patients who had the final diagnosis of RLS, 25.7 % had referred to a physician for the symptoms and 22.2 % got the correct diagnosis. The RLS prevalence in Ankara was somewhere between Western and Far East countries compatible with the geographical location. Diagnostic criteria may not be fully predictive when applied by non-physician pollsters. Physician's probability of correctly diagnosing RLS is still low.