The test-retest reliability of self-rated health is analysed and compared with the reliability of health questions phrased more as well as less precisely. Differences in reliability between men and women and between age groups are also assessed. The study is based on 204 and 409 re-interviews from the 1991 Swedish Level of Living Survey and the 1989 Survey of Living Conditions respectively. The results show that the reliability of self-rated health is as good as or even better than that of most of the more specific questions. Only an indicator of high blood pressure showed significantly higher reliability. The reliability of self-rated health is good in all subgroups studied, and is even excellent among older men. It is concluded that the good overall reliability of self-rated health found in this study is in line with previous results concerning the validity of people's assessments of their general health as well as results concerning the basis upon which they make these judgements.