We studied the test-retest reproducibility of melanoma risk factors, including the use of sunscreen and sun protection factor (SPF), in a self-administered exposure follow-up questionnaire from the Norwegian Women and Cancer Study, a large national population-based cohort. In 2002, a random sample of 2000 women (46-75 years) received the questionnaire twice, about 3 months a part (response 75%). Kappa (kappa) was 0.77 for freckling when sunbathing [95% confidence interval (CI), 0.74-0.81]. Weighted kappa, kappaw, for sunbathing vacations to southern latitudes and solarium use last 5 years was 0.71 (95% CI, 0.68-0.74) and 0.70 (95% CI, 0.67-0.73), respectively. Reproducibility was also good for sunscreen use (yes/no) on specific occasions (0.64< or =kappa< or =0.74) and the corresponding SPF. Spearman's correlation coefficient (r(s)) for SPF on sunbathing vacations to southern latitudes was 0.73 (95% CI, 0.69-0.77) for today and 0.71 (95% CI, 0.66-0.76) for 10 years ago. For the eight most common sunscreen brands, reproducibility was lower for use (yes/no) (0.31< or =kappa< or =0.60) than for SPF (0.38< or =r(s)< or =0.87). The frequency of sunburn and sunbathing vacations in Norway or outside southern latitudes had fair reproducibility (kappaw was 0.49 and 0.47, respectively). Other studies have also found it challenging to measure sunburn. This study was larger than previous studies, permitting subgroup analyses. In conclusion, the overall reproducibility of the questionnaire was acceptable and not affected by age, education or skin color. In particular, our study has added new knowledge about the reproducibility of sunscreen use and SPF.