Abstract We constructed a varve data base (VDB) to study the worldwide distribution of varved sediment records and the fidelity of their varve chronologies. Clusters of lakes with varved sediments exist in central North America, in the Canadian Arctic, and in northern and central Europe. Among the 108 sites included in the VDB, varved records are typically 200–500 cm long and cover a period of 1000–2000 years. Their varve chronologies are often based on counting of annual layers from fresh sediment surfaces and photographs or from epoxy embedded sediment blocks and thin sections. The VDB indicates that chronological errors associated with counting sedimentary varves fall generally between 1 and 3% but only 57% of published records are providing quantitative error estimations for their chronologies. With a careful documentation of varves and by applying radiometric dating methods to surface sediments as well as using historical events as time markers, there is the potential of having centennial-long varved records with a very precise age–depth control. However, it is unrealistic to expect errors significantly below ±1% from several millennia long sections. We found no indication that varve chronologies would have been substantially more accurate and precise in some parts of the world than in others or statistically dependent on varve thickness or temporal extension of a varved sequence. According to the VDB, close to 90% of the published varve chronologies have been cross-checked with some independent dating methods.