Our study has several limitations. First, we diagnosed RLS based on participants’ reports according to IRLSSG criteria and some conditions similar to RLS may be included. It is known that these criteria can mimic other conditions including nocturnal leg cramps, anxiety disorders, akathisia, meralgia paresthetica, and peripheral neuropathies [44]. To minimise the accidental inclusion of these conditions, a new diagnostic questionnaire was recently proposed [53]. It showed a high specificity but has not yet been used in an epidemiological study. Second, we did not assess the severity of RLS due to limits on the questionnaire length [54]. An attempt to model a quantitative correlation between migraine and RLS may add better insights to the association between these 2 disorders. As such, further study including the RLS severity scale would be needed to better understanding this association. Third, although this is a population-based study with a low sampling error, its statistical power for examining subgroups was limited. Thus, some results might have not reached statistical significance merely because of the limited sample numbers, especially for small groups based upon the age-band. Fourth, we did not thoroughly investigate the secondary causes of headache because this is difficult to document with the questionnaire method used in this population study. Although most of recurrent headache sufferers in general population were considered to have primary headaches rather than secondary headaches, some secondary headaches sufferers might be classified as having primary headaches [55]. The 1-year prevalences of all headaches and migraine were similar to previous studies in Korean and other Asian countries [3, 6, 7, 34, 56].