Carbon sequestration in agricultural soils through improved management practices has been identified as one option that helps mitigate carbon dioxide emissions. The aim of this study was to model national-scale soil carbon dynamics in the US. One main component was estimating the crop residue carbon inputs to agricultural soils using national crop yield and acreage data, for the period 1982- 1997; one difficulty in using crop information in this estimation was the lack of a database with complete information on county level yields and acreages. We found that there are gaps or missing data in the annually reported crop data by the National Agricultural Statistics Service (NASS). We filled some of the gaps in NASS yield data using regression analyses with the data produced by the Census of Agriculture (Ag Census) that are reported every five years. The remaining gaps in the yield data were filled using mixed linear effect models that incorporated county level weather variables (mean monthly summer temperature, precipitation, potential evapotranspiration, etc.) in relation to different agro-eco zones of the US. Gaps in NASS acreage data were filled using Ag Census acreage data, and mixed linear effect models using several economic and environmental variables. Thus complete crop yield and acreage databases for the major crops in the US were created, and carbon inputs to agricultural soils from the residues of those crops were estimated.