An efficient way to detect spatial differences in crop and soil conditions at field scale is through image-based remote sensing. We report on three studies investigating the estimation of soil properties using bare-soil images. First, the ability of image data to estimate soil chemical property levels in a central Korean rice plain area was investigated. Quickbird and Orbview satellite images were obtained in October 2004, April 2005, and May 2005 during the non-growing season for a rice plain in the Dangjin-gun area where Low Humic-Gley soils (fine silty, mixed, nonacid, mesic family of Fluvaquentic Endoaquepts) predominate. Soil samples were collected at 145 irregular grid points from about 500 ha of rice plain and were analyzed in the lab. Digital numbers (DN) obtanied in each band (blue, green, red, and near infrared) were related to soil chemical properties, including soil organic matter, cation exchange capacity (CEC), exchangeable cations (Ca, Mg, K), pH, and available phosphate. Organic matter was negatively correlated to DN in all bands of the image taken in April 2005, when the rice paddy fields were tilled and irrigated for transplanting. Other than organic matter and CEC, soil chemical properties were not significantly correlated to the spectral signatures. Second, a hyperspectral aerial image sensor was evaluated for estimating soil chemical properties, soil texture, and bulk soil electrical conductivity (ECa). Airborne bare soil images were acquired using a prism grating pushbroom scanner (RDACSH3) in April 2000, May 2001, and June 2002 for a central Missouri experimental field (fine, smectitic, mesic aeric Vertic Aqualfs) in a minimum-tillage corn-soybean rotation. Data were converted to reflectance using chemically-treated reference tarps with known reflectance levels. Geometric distortion of the pushbroom sensor images was corrected with a rubber sheeting transformation. A 5-m image pixel size was used, based on analysis of short-range variations in five sub-field areas. Blue wavelengths showed the strongest correlations with soil properties. Clay, organic matter, exchangeable cations (Ca, Mg, K), CEC, and ECa were negatively correlated with reflectance. Bare soil images obtained under dry soil conditions (2000 and 2002) were more useful than moist soil data for estimating soil chemical properties and ECa. Moist soil reflectance (2001) was more strongly related to soil texture than was dry soil data. Third, spectral reflectance data obtained during a two-day period with three different image sensors – two airborne hyperspectral sensors (RDACSH3 and AISA) and a satellite sensor (Quickbird) – were related to soil properties to compare image quality and performance. Random noise present in the RDACSH3 image was higher than that of AISA or Quickbird images in terms of the ratio of mean to standard deviation in five subfield areas of each image. Regardless of sensor type, blue wavelengths were most informative for all ground measured soil properties.