The impact of disease on the ability of a person to perform work should be part of an economic evaluation when a societal viewpoint is used for the analysis. This impact is reflected by calculating productivity costs. Measurement of these costs is often performed retrospectively. The purpose of our study was to study precision and accuracy of a retrospective self-administered questionnaire on sick leave. Employees of a company were asked to indicate the number of days absent from work due to illness during the past 2 weeks, 4 weeks, 2 months, 6 months, and the past 12 months. The percentage of respondents with an absolute difference of a maximum of respectively 0, 1, 2, 3, 4, 5, 6, 7, 8, and 9 or more days between reported, and company-registered absence due to illness was determined. Besides this, the proportional difference was calculated. A systematic difference was tested with a signed rank test. Of the reported data, 95% matched the registered data perfectly when the recall period was limited to 2 and 4 weeks. This percentage decreased to 87%, 57%, and 51% for 2 months, 6 months, and 12 months. The weighted mean proportional differences for the recall periods were respectively 32.9, 35.2, 45.3, 34.9, and 113.6%. No systematic positive or negative difference was found between registered and reported sick leave. The results suggest that the recall period for retrospective measurement of sick leave is limited according to the precision level, which seems to be appropriate for the subject and the purpose of the study. We recommend using a recall period of no more than 2 months.