The use of questionnaires in epidemiological studies needs more methodological research. The time and effort spent on questionnaire design is often limited. Studies on the construction of questionnaires could lead to a higher quality of data, enhanced comparability and improved credibility of epidemiological findings. The aim of the present study was to examine the test-retest reliability of some common items measuring sun-related variables. A sample of 52 female Swedish nurses attending a postgraduate course in research methodology was chosen. They completed a questionnaire on two occasions spaced 3 weeks apart in the winter of 2000. When the results were analysed, items on sun-related behaviours and sunbed use were sufficiently reliable. The items on skin type and sunburn showed moderate stability. Stage of change related to sunlight exposure, items measuring beliefs about sunbathing along with items assessing self-efficacy and risk perception with regards to sunbathing showed lower stability. The results showed that many essential items concerning exposure to ultraviolet radiation and sunburns were sufficiently stable, but other items were less stable and could be improved upon; suggestions for improving these items are presented. The study illustrates the value of reliability testing in the process of item construction. Using methodological studies to improve the reliability and validity of data is an important step toward higher standards for questionnaire surveys.