Near infrared spectroscopy (NIRS) is an effective technique for simultaneously measuring several soil properties including soil organic carbon, total nitrogen, moisture, and cation exchange capacity. However, developing robust calibration models for predicting soil properties from spectral measurements is a challenge because of the non-linear response of NIR spectra to soil property variations. Regional calibrations may be a good approach to meet accuracy specifications while providing a practical scale of applicability. This study investigates the use of a set of regional calibrations to measure soil properties in Bremer and Black Hawk counties in north central Iowa. Spectral measurements were acquired in nine fields using an on-the-go NIR spectrophotometer. Calibrations were made using partial least squares regression (PLS) with various spectral pre-treatments. Leave-one-out and leave-one-field-out cross validation are used to examine robustness of the regional calibration models. Finally, calibrations are applied to the spectral data to create field maps for several different soil properties.