Using lake sediments to infer past total mercury and methylmercury loading to the environment requires that diagenetic processes within the sediment do not significantly affect the concentrations or net accumulation rates of the mercury species. Because carbon is lost during early sediment diagenesis, the close link between carbon and mercury raises the question of how reliable lake sediments are as archives of total mercury and methylmercury loading. In this study we used a series of freeze cores taken in a lake with varved (annually laminated) sediment to assess the stability of total mercury and methylmercury over time. By tracking material deposited in specific years in cores collected in different years, we found that despite a 20--25% loss of carbon in the first 10--15 years, there was no apparent loss of total mercury over time; hence, lake sediments can be considered as reliable archives. However, over the first 5--8 years after sedimentation, about 30--40% of the methylmercury was lost (a decrease of 0.025--0.030 microg MeHg m(-2) yr(-1)), suggesting that sediment profiles showing increasing methylmercury concentrations toward the sediment surface are in large part an artifact of diagenetic processes (net demethylation), rather than a record of changes in methylmercury loading.