The consolidated wave 1 of the Belgian Generations and Gender Survey (GGS), which is currently accessible on the Generations and Gender Programme platform (GGP-i), has errors in dates of union formation. The Belgian GGS use a broad definition of partnership history, collecting data for all non-coresident and coresident unions. Dates of the start of the relationship and the start of cohabitation were both covered in the survey. The harmonization of the data for all national surveys has led to two errors for Belgium. First, all “living apart together” partnerships are included in the database but not reported as such. Second, dates of entry into cohabitation for past unions actually refer to dates of the start of the relationship. In this note, I describe this problem and its consequences for cross-country comparison. Using the raw Belgian GGS dataset, I also propose a new version of the consolidated survey in order to revise partnership histories for future cross-national comparative analyses.