Accurate field-scale maps of total soil carbon are important not only for defining site-specific management zones, but also for reliable assessment of carbon sequestration potential. Highly variable soil and topographical attributes of glacial till soils cause difficulties in mapping soil carbon with limited number of soil samples. The first objective of this study was to study performance of the newly developed on-the-go near infrared spectroscopy (NIRS) system for predicting total soil carbon. The second objective was to use NIRS and topographical data to create high quality soil carbon map. The studied field was about 50 ha in size and located in Kalamazoo County. The predominant soil is Kalamazoo loam (fine-loamy, mixed, mesic Typic Hapludalfs). NIRS spectra were collected along 22 north-south transects separated by distance of 25 m, the distance between the measurements within the transect was 5 m. Eighty-five soil samples were collected at the path of NIRS from depth 0-10 cm. Principal component regression was used to relate NIRS spectra to measured soil carbon. Regression coefficient between measured and predicted carbon values using NIRS data was equal to 0.70. When topographical attributes, such as elevation and curvature were included into regression along with NIRS, the regression coefficient improved to 0.8. The results indicated that combining the NIRS measurements with topography allowed us to improve accuracy of the total carbon map in glacial till soils of Michigan.