By coupling the SPOT-5 multi-spectral RS images, ground-spectrum, and field measured data of different winter wheat ecological zones, a pure pixel spectrum extraction method was developed based on spectral response function and pixel unmixed, and the quantitative relationships between leaf nitrogen accumulation (LNA) and simulated, measured, and pure pixel spectra were analyzed. The estimation accuracy for LNA was in the sequence of simulated pixel spectra > pure pixel spectra > measured pixel spectra. However, the LNA monitoring model based on simulated pixel spectra couldn't be extrapolated directly to spatial level. The results of model verification also indicated that the monitoring model based on pure pixel spectra performed well in two different wheat ecological zones. Therefore, the pure pixel spectrum extraction method could be applied to other varied and remotely sensed data with different spatial and spectral resolutions by making use of the merits of ground- and space- remote sensing simultaneously, which provided a technological basis for estimating winter wheat nitrogen status in regional scale.