Modern agricultural practices are increasingly making use of conservation (reduced- and no-till) methods, in order to minimize soil erosion and increase soil organic carbon (SOC) content. These methods result in increased crop residue cover after planting when compared to conventional tillage methods. Traditional methods for residue cover estimation are time and manpower intensive. As such, remote sensing methods offer promise for rapid and inexpensive residue cover estimation over large areas of land. However, surface spectral properties vary with soil composition (mineralogy and SOC type and content), and crop residue cover spectral properties are affected by residue type, age, water content, and encrusting by soil and mold. In this work, we evaluate several remote sensing indices for residue cover estimation, including continuum-removal methods such as the Cellulose Absorption Index (CAI) and the ASTER Lignin-Cellulose Absorption Index (LCA), and normalized difference methods such as NDSVI, NDTI, NDI5, and NDI7. Images were acquired from air- and spaceborne hyperspectral sensors over research sites over fields in the Indiana, Illinois, and Maryland. Of these indices, CAI fared best (r²=0.71 for Indiana, r²= 0.76 for Illinois), followed by LCA, with Landsat TM indices faring the worst. This was expected as residues and soils are best differentiated using narrow spectral bands from 1950 nm to 2450 nm, rather than using the broad TM band reflectances. Our results showed that when soil spectral data and crop type information were used for calibration for an inverse solution, residue cover estimation improved significantly compared to uncalibrated regression analyses.