The content on Facebook pages is obviously subject to changes. However, there is often a considerable time lag between the sampling period and the collection date in studies on public Facebook communication. The present evaluation study aims to assess whether - and if so, how - problematic this common practice is in terms of unavailable content. I investigated whether a sample of content items, which were posted to 408 German political Facebook pages in March 2017, could be retrospectively collected in June 2017, September 2017, and March 2018. 27% of the 132,068 test items were no longer accessible after 12 months. The deletion of complete pages was rare, but some prominent pages were not available retrospectively. Posts by the pages themselves were far more likely to be available compared to content, which was posted by private accounts to the pages. There were only small differences between the political parties, but there was substantial variation between the distinct pages. I conclude with a discussion of implications for single research projects (applying timely data collection whenever possible) and for communication research and related disciplines (considering institutionalized data collection efforts).