Since 1850, land use has been the second largest global net source of CO₂ emissions. Estimating carbon loss due to historical logging and agricultural clearing, and subsequent sequestration from forest regrowth, is thus of great importance to balancing carbon budgets. Using historical survey records, we assessed fine-resolution patterns of above-ground live forest biomass (AGB) across Wisconsin, USA (145 000 km²) in the: (1) mid-1800s, prior to Euro-American settlement; (2) 1930s, following widespread logging and clearing; and (3) 2000s, following forest regrowth. We combined two sets of historical records, the U.S. Public Land Survey (mid-1800s) and Wisconsin Land Economic Inventory (1930s) with U.S. Forest Service FIA data to estimate biomass, and used Monte Carlo simulation to assess uncertainty due to bias in the historical records and error in the allometric biomass equations. Our results show that AGB declined 3.5-fold from the mid-1800s to 1930s, and despite forest regrowth is still only 60% of its former level. Spatial variability in AGB declined due to deforestation in the north and forest ingrowth in the south. Deciduous forests currently hold a much larger proportion of total AGB than in the mid-1800s. Monte Carlo simulation was a useful method for estimating AGB from historical survey records, and showed that the greatest source of uncertainty resulted from uncertainties in historical tree diameter distributions. These estimates could be used to set future sequestration goals in Wisconsin, although AGB recovery will likely be limited by the loss of historical forests to agriculture, and harvesting practices in extant forests